Mining transcriptomic data to identify Saccharomyces cerevisiae signatures related to improved and repressed ethanol production under fermentation

Sima Sazegari; Ali Niazi; Zahra Zinati; Mohammad Hadi Eskandari

doi:10.1371/journal.pone.0259476

Abstract

Saccharomyces cerevisiae is known for its outstanding ability to produce ethanol in industry. Underlying the dynamics of gene expression in S. cerevisiae in response to fermentation could provide informative results, required for the establishment of any ethanol production improvement program. Thus, representing a new approach, this study was conducted to identify the discriminative genes between improved and repressed ethanol production as well as clarifying the molecular responses to this process through mining the transcriptomic data. The significant differential expression probe sets were extracted from available microarray datasets related to yeast fermentation performance. To identify the most effective probe sets contributing to discriminate ethanol content, 11 machine learning algorithms from RapidMiner were employed. Further analysis including pathway enrichment and regulatory analysis were performed on discriminative probe sets. Besides, the decision tree models were constructed, the performance of each model was evaluated and the roots were identified. Based on the results, 171 probe sets were identified by at least 5 attribute weighting algorithms (AWAs) and 17 roots were recognized with 100% performance Some of the top ranked presets were found to be involved in carbohydrate metabolism, oxidative phosphorylation, and ethanol fermentation. Principal component analysis (PCA) and heatmap clustering validated the top-ranked selective probe sets. In addition, the top-ranked genes were validated based on GSE78759 and GSE5185 dataset. From all discriminative probe sets, OLI1 and CYC3 were identified as the roots with the best performance, demonstrated by the most weighting algorithms and linked to top two significant enriched pathways including porphyrin biosynthesis and oxidative phosphorylation. ADH5 and PDA1 were also recognized as differential top-ranked genes that contribute to ethanol production. According to the regulatory clustering analysis, Tup1 has a significant effect on the top-ranked target genes CYC3 and ADH5 genes. This study provides a basic understanding of the S. cerevisiae cell molecular mechanism and responses to two different medium conditions (Mg²⁺ and Cu²⁺) during the fermentation process.

Citation: Sazegari S, Niazi A, Zinati Z, Eskandari MH (2022) Mining transcriptomic data to identify Saccharomyces cerevisiae signatures related to improved and repressed ethanol production under fermentation. PLoS ONE 17(7): e0259476. https://doi.org/10.1371/journal.pone.0259476

Editor: Kandasamy Ulaganathan, Osmania University, INDIA

Received: October 13, 2021; Accepted: July 12, 2022; Published: July 26, 2022

Copyright: © 2022 Sazegari et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data are within the paper and its Supporting Information files.

Funding: The author(s) received no specific funding for this work.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Saccharomyces cerevisiae is used as one of the main microorganisms for bio-ethanol production in research and industry. In addition to high ethanol production potential, stability for anaerobic fermentation and low pH tolerance, facilitates its use in industry for ethanol production [1]. Among industrial S. cerevisiae strains used in fermentation industry, JP1 is one of the dominants exhibits high fermentation rate due to high temperature tolerance and low pH stability [2]. Since many of S. cerevisiae genes have been functionally annotated and characterized, genetic manipulation of this organism is well developed [3–5]. Several research have been conducted on the S. cerevisiae metabolic engineering to generate efficient ethanol producing strains [6, 7]. Suji et al [8], for example used the PHO13 deletion in conjunction with LAD1 and ALX1 heterologous expression to improve S. cerevisiae for arabinose consumption, resulting in a 3.5-fold increase in specific ethanol productivity. Although the manipulation of S. cerevisiae strains for higher ethanol production is highly remarked by researchers, a few studies have been conducted regarding the molecular modulations during enhanced and repressed ethanol fermentation. Such research, on the other hand, is critical for improving the efficiency of industrial applications. De Souza et al. [9] reported the presence of stress-response and energy-related genes under optimized fermentation condition supplemented by Mg²⁺ [9]. Exploring for genes that are linked to ethanol tolerance in Saccharomyces. C via transcriptomic analysis revealed that glucose metabolism and energy related genes are induced by ethanol stress. Moreover, some processes such as oxidative phosphorylation and cellular respiration shown to be affected by ethanol stress [10]. Identifying the molecular basis and dynamics of gene expression profiles related to yeast response in improved bioethanol production conditions is critical for developing new manipulated strains with increased ethanol yield. It also shed light on the mechanisms that yeast uses to improve production. Metal supplements significantly affect the ethanol production associated metabolic pathways in yeast. Among these, zinc, magnesium, manganese, and copper have been extensively studied and shown to have regulatory effects on ethanol production [11–13]. Mg²⁺ ion is involved in phosphorylation, DNA and protein synthesis, cell membrane rigidity and proliferation, along with the potential to increase ethanol accumulation through fermentation [9, 14]. Furthermore, Mg²⁺ may improve the S. cerevisiae tolerance to high ethanol concentration during glucose and xylose fermentation [15, 16]. S. cerevisiae medium supplementation with Mg²⁺resulted in a 29% increase in ethanol production through regulating cell wall and membrane related genes expression [16]. Copper ion (Cu²⁺) is also known as a critical element for yeast biological functions and involves in a variety of metabolic pathways. For instance, cytochrome c oxidase, a component of oxidative phosphorylation, and superoxide dismutase as an important enzyme contribute to stress regulation, are dependent on Cu²⁺ [17]. Copper stress, on the other hand, caused by an excess of copper, can result in reactive oxygen species (ROS) generation and DNA damage. At high concentrations, it also has a negative impact on cell membrane stability and enzyme activity. [18]. A high copper concentration (1.5 mM) inhibited cell growth, glucose and fructose consumption during fermentation by S. cerevisiae [19]. However, few studies have been conducted to investigate the effect of Cu²⁺ on the molecular physiology and fermentation ability of S. cerevisiae cell. Despite their lack of research, Mg and Cu have the potential to modulate the gene expression network involved in the fermentation process.

It would be possible to identify the critical genes and clarify the molecular mechanisms involved in the ethanol production process using bioinformatics-based analysis of the S. cerevisiae expression dataset. Computational approaches for identifying key genes involved in the fermentation process could elucidate the transcriptomic dynamics of yeast ethanol fermentation and reveal expression signatures contribute to improved production. RapidMiner is one of the most useful and comprehensive mining tools in data science [20]. Different gene selection algorithms, such as Information Gain, Information Gain Ratio, rule induction, support vector machine (SVM), and PCA, are widely used in gene expression analysis using RapidMiner. Machine learning algorithms, both supervised and unsupervised models, are widely used in gene expression data analysis and gene identification [21, 22]. Cheng et al [23] used RapidMiner to preform four machine learning weighting models on gene expression datasets related to Huntington’s disease, including decision tree, rule induction, random forest, and generalized linear algorithms, in order to identify contributing genes to this disorder. Huang et al. [24] conducted Modelling with network features and machine learning approach to identify the functional features that affect the S. cerevisiae longevity. Mitochondria and cell cycle were found to have determinant role in yeast life span. However, to the best of the authors’ knowledge, no machine learning tools via RapidMiner have been utilized to investigate the molecular basis of metabolic pathways in S. cerevisiae. Valuable publicly available data related to S. cerevisiae genome-wide expression experiments could be used for functional genomic analysis through machine learning. Machine learning algorithms’ discriminative ability aids in revealing the unravelling biological process from microarray data sets [20]. In light of the availability of such useful primary data sets and the potential of RapidMiner as an efficient tool for biological data analysis, we used available microarray expression dataset related to S. cerevisiae supplemented with Copper and Magnesium metal components under fermentation to investigate the underlying molecular basis of fermentation used by S. cerevisiae. The aim of this study was to identify the critical genes discriminate the improved (Mg²⁺ treatment) and low ethanol production (Cu²⁺ treatment at toxic concentration) and elucidate the transcriptomic response of S. cerevisiae under these two conditions. S. cerevisiae transcriptome analysis using data mining and machine learning by both supervised and unsupervised models was used in this study as a novel approach to identify the underlying gene regulation mechanisms that can be used to optimize fermentation performance.

Materials and methods

Data collection

For this study available microarray datasets related to yeast fermentation performance under Mg²⁺ (500 mg/L) or Cu²⁺ (1 mg/L) supplementation was used. Microarray data of the industrial yeast S. cerevisiae JP1 strain downloaded from the GEO repository of the NCBI database (GEO number: GSE75803) was used. To meet the research objective, the probe sets with significant differential expression (concomitant Adj. p < 0.05 and B ≥ 3) were chosen for this study.

Data cleaning

We used RapidMiner software (RapidMiner Studio 7.6) [25] to enter the 6300-differential expressed probe sets as numerical features, as well as high and low bioethanol as class features. For better processing, inefficient or redundant probe sets with less than or equal to a given standard deviation (SD) threshold (0.1), as well as correlated probe sets (correlation ≥ 0.95), were carefully removed from the dataset. The resulting list, which only contained efficient probe sets, was designated as the Final Cleaned (FCdb) database.

Attribute weighting algorithms

Eleven attribute weighting algorithms with cut-off ≥ 0.7 were used in the FCdb to identify the most effective probe sets contributing to discriminate ethanol content. Weights close to 1 indicate that a specific probe set in ethanol content is more important. The main probe sets were those determined by the majority of AWAs (intersection of the weighing method). The attribute weighting algorithms used in this investigation, as well as the statistical background description for each one, are as follows (RapidMiner Studio 7.6):

Weight by information gain and information gain ratio

This algorithm is a well-established superior method for gene selection in microarray data analysis [26, 27]. In this method, the attributes (probe sets) are weighted according to their class label (high or low ethanol production).

Weight by rule

Based on a single rule and the relationship between attributes (genes) and considering the errors, the weight of each attribute is measured through rule algorithm [28] and is used as a selective method for microarray analysis.

Weight by deviation, weight by correlation and weight by chi squared statistic

The standard deviation of attributes is used as a weighting parameter in the deviation weighting method. The correlation method, on the other hand, weighs the label attributes based on the correlation. In addition, for labeling the attributes, we used the Chi Squared Statistic weighting algorithm, which takes the Chi squared into account.

Weight by Gini index and weight by uncertainty

Due to the label attribute in this model, the weight of attributes is determined by measuring the Gini coefficient as an inequality index of sample data. According to each attribute, the lower the Gini index of the attribute, the more equal dispersion among attributes is considered. The weight for uncertainty model, on the other hand, is determined using the symmetrical uncertainty due to the class attribute.

Weight by relief

This model is one of the most reliable algorithms for weighting genes because it is based on the determination of values between probe sets of the same and different classes in a short distance.

Weight by SVM

SVM is one of the most powerful classification models for gene expression analysis [21]. The SVM method weighs attributes using the coefficients of the normal vector of a linear SVM.

Weight by PCA

This model performs attribute weighting due to the class attribute based on the component number parameter of PCA and the value of the components.

Decision tree models

Eleven new datasets were generated using the entire probe sets with weight >0.70. They were annotated based on the models used for attribute weighting (Relief, Information gain, Uncertainty, Information gain ratio, Chi Squared, Rule, Correlation, Deviation, SVM and PCA, Gini index). Random Tree, Decision Tree, Decision Stump, and Random Forest were the tree induction models used for 12 datasets (FCdb and 11 datasets produced by specific weighting algorithms). Each model had four criteria (Gini Index, Gain Ratio Information Gain, and Accuracy). We used a ten-fold validation algorithm with appropriate sampling to create trees with RapidMiner. The performance of the model was evaluated and used to compare various models based on the accuracy of each model in identifying the target variable (high and low bioethanol content) and according to the attribute variables (normal expression of the probe set). Performance is expressed as a measure of model accuracy in this case. We calculated the accuracy by dividing the number of correct predictions by the total number of samples. The value of the attribute accuracy that is expected to be the same as the value of the labeled attribute is referred to as the correct prediction. These models were used with a minimum gain of 0.1 to obtain a split and a maximum tree depth of 20. For pruning 0.25 confidence level was considered with a pessimistic error calculation.

Unsupervised analysis of the top ranked probe sets derived by supervised AWAs

Unsupervised principal component analysis (PCA) and hierarchical clustering heatmap were used to evaluate the power of top-ranked probe sets which differentiate the fermentation under different supplementation treatments. For unsupervised analysis, a web-based tool Clustvis (https://biit.cs.ut.ee/clustvis/) was used [29]. The PCA analysis was carried out in the PCA Methods R package using unit variance scaling on rows and Singular Value Decomposition (SVD) with the imputation method. The clustering heatmap was created with the pheatmap R package (version 0.7.7). The clustering heatmap was constructed using correlation, Pearson correlation subtracted from 1, and the average distance of all possible pairs [29].

Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis

The pathway enrichment analysis was carried out using YeastEnrichr (https://maayanlab.cloud/YeastEnrichr/) [30, 31]. The biochemical pathways related to key probe sets were identified using the KEGG2019 database. Pathways with p-value < 0.1 were considered significant.

Exploring for transcription factors among top-ranked genes and regulator cluster analysis

We used yeastract database (http://www.yeastract.com/formrankbyhomotf.php) to identify transcription factors (TFs) among the 171 probe sets identified by at least 5 attribute weighting algorithms [32]. The TFs and their target genes were identified using this tool based on DNA binding sites and expression evidence. Furthermore, we used the regulator DB database (http://wyrickbioinfo2.smb.wsu.edu/cgi-bin/RegulatorDB/cgi/home.pl) to run regulator cluster diagram to determine the regulatory effect of the identified TFs on the target genes [33, 34]. It provides data on mutant regulator expression for selected regulators and target genes.

Independent validation of top-ranked genes using different microarray expression datasets

For independent validation of the top-ranked genes determined by supervised attribute weighting models, impartial samples of high or low ethanol production from microarray experiment with GEO accession of GSE78759 and GSE5185 [35, 36] were selected. The original experiments regarding GSE78759 had been designed to investigate the transcriptomic changes occurred in two ethanol-tolerant strains with maximum ethanol production rate, selected through evolutionary engineering, in comparison with their reference as low ethanol producing strains. The other dataset, GSE5185, was related to improved Ethanol producing mutant generated by transcription machinery engineering compared with the wild strain as lower ethanol producing. We used 5 samples including GSM2075761, GSM2075762, GSM2075763, GSM2075764, GSM2075765 as high ethanol producing strains and 3 samples including GSM2075758, GSM2075759, and GSM2075760 as low ethanol producing from the first dataset as well as GSM116819 and GSM116820 as high ethanol producing and GSM116815, GSM116816, GSM116817 as low ethanol samples from the second datasets. The 171 probe sets which were identified by AWAs were selected among the samples and the unsupervised principal component analysis (PCA) were used to evaluate the power of top-ranked probe sets to separate the low and high samples. Besides, validation of the top-ranked genes was implemented using a leave-one-out cross-validation (LOOCV) on expression values of the top-ranked genes. Attribute (expression values of 171 probe sets) were chosen using the LOOCV mode in which expression matrix of one sample discarded for the test and the remaining parts for training. Discrimination was performed based on Logistic function to model a binary classification of samples according to their gene expression values.

Results

Ranking probe sets by AWAs

After cleaning 6300 probe sets by RapidMiner, we obtained 1813 probe sets. Eleven AWAs were used to identify informative probe sets. Following AWAs analysis, 171 probe sets were identified by at least 5 attribute weighting algorithms (S1A File). Furthermore, there were distinct probe sets classified by at least five algorithms that respond discriminatively to supplement treatment and/or are particularly related to ethanol production during fermentation. Sheet B of the S1 File contains the probe sets as well as the AWAs used to identify the probe sets. Some of the informative probe sets were recognized to be involved in carbohydrate metabolism, TCA cycle, oxidative phosphorylation, and ethanol fermentation while others were related to stress responses, cell membrane structure, and cell growth which could be indirectly effective in ethanol production. Some of the top informative genes are presented in Table 1. However, it should be noted that some of discriminative probe sets which respond to Mg²⁺ and Cu²⁺ supplementation may not be required in or influence the ethanol production process.

Download:

Table 1. Some of the informative probe sets identified by at least four AWAs.

https://doi.org/10.1371/journal.pone.0259476.t001

Decision tree models

Sixteen different decision tree models were applied on twelve datasets (eleven new datasets produced by trimming FCdb based on a weight ≥ 0.7 given by each attribute weighting approaches, as well as FCdb) were used to achieve pattern recognition between important genes as well as with the genes with the highest distinguishing power. The performance of each model was evaluated and the lowest and highest performances were 0% and 100%, respectively (S1C and S1D File). The entire model trees were characterized with root node, branches and leaf nodes (class label; high or low ethanol). The topmost node in the tree is the root node. Genes in the root of decision tree have stronger effects on determining the general data pattern than genes in the tree branches. There were 17 probe sets with 100% performance in the roots of decision tree models Table 2. The root list contains OLI1, CYS3, HTB2, CGR1, CYC3, RPS13, OST4 and COX1 genes with known biological process. Albeit probably contributed to ethanol fermentation, the biological processes of the other genes determined as the roots of decision trees are not known yet.

Download:

Table 2. Decision tree models roots identified as exhibited 100% performance.

https://doi.org/10.1371/journal.pone.0259476.t002

Unsupervised analysis

As a complementary confirmation, the 171 top ranked probe sets were validated using PCA and hierarchical clustering heatmap, identified with supervised attribute weighting models. According to the results, the 171 significant probe sets could accurately differentiate between two different fermentation conditions, thus confirming the significance and accuracy of the identified probe sets (Fig 1). In particular, the captured variances with the first two components on all recognized 6031 probe sets and informative 171 probe sets were up to 75% and 50%, respectively. Furthermore, it could efficiently separate informative 171 probe sets under Cu²⁺ or Mg²⁺ supplementation in the hierarchical clustering heat map, (Fig 1). In total, 64 probe sets were up and down regulated by Mg²⁺ and Cu²⁺, while 108 probe sets were up and down regulated through Cu and Mg supplementation, respectively (Fig 2).

Download:

Fig 1. Two-dimensional plot related to the first two principal components.

GSM1968101, GSM1968110, GSM1968100 and GSM1968108 are samples related to Mg²⁺ supplementation. GSM1968106, GSM1968114, GSM1968103 and GSM1968112 are samples related to Cu²⁺ supplementation.

https://doi.org/10.1371/journal.pone.0259476.g001

Download:

Fig 2. The heatmap related to 171 probe sets which were recognized by at least 5 attribute weighting algorithms (AWAs).

Each row corresponds to the different samples including Mg²⁺ (high ethanol production) and Cu²⁺ supplementation (repressed ethanol production). Columns exhibits hierarchically clustered probe sets. The normalized intensity expressions of probe sets were shown as a color scale. The up and down-expression levels were represented as red and blue scales, respectively.

https://doi.org/10.1371/journal.pone.0259476.g002

Pathway enrichment analysis of genes

A total number of 21significant terms were recognized and enriched in ranked probe sets (adjusted p-value <0.1). Some probe sets including PDA1 and CYC3 participated in multiple pathways. Significant related enriched pathways such as Porphyrin metabolism, Oxidative phosphorylation, Glycolysis, Amino sugar and nucleotide sugar metabolism, Cell cycle, MAPK signaling pathway and Citrate cycle (TCA cycle) were identified using the KEGG enrichment analysis. The porphyrin metabolism pathway shown to be the most significant defined by probe sets list. Confirmed by Qi et al. [37], the porphyrin biosynthesis is reported to be associated with ethanol fermentation in yeast through controlling ROS level under stress condition. On the other hand, the topmost genes, including COX1 and OLI1, that were ranked by the highest number of AWAs involved in Oxidative phosphorylation pathway. This suggest that this pathway is crucially regulated by S. cerevisiae Cu²⁺ and Mg²⁺ supplementation experiment. The enriched pathways and the related genes are presented in Table 3.

Download:

Table 3. KEGG enrichment analysis of 171 probe sets.

The significant pathways with adjusted p-value < 0.1 are represented.

https://doi.org/10.1371/journal.pone.0259476.t003

Identification of transcription factors and their targets

Among the 171 informative probe sets identified by yeastract analysis were seven transcription factors: YGR067C, HAP4, NRG2, TUP1, TOS8, MSN4, and PDC2. Surprisingly, the targets of the identified transcription factors were discovered among the 171 genes identified by RapidMiner analysis and ranked by at least 5 algorithms (S1E File). These findings support the AWAs’ ability to correctly identify top-ranked probe sets. Furthermore, regulator clustering related to TFs and their targets (both ranked by at least eight algorithms) was performed to demonstrate the effect of top-ranked TFs on top-ranked target genes based on the transcription factors mutants. The results showed that Hap4p, Tup1, and TOS8 mutants resulted in different ratios of up and down-regulation of target genes (Fig 3). Although Hap4 and TOS8 resulted in down or up regulation of target genes, their effect on none of target genes was significant. According to the findings, the Tup1 transcription factor has the greatest impact on the target genes expression. The Tup1 knocked out mutant significantly induce the expression of CYC3 (YAL039C), while causing highest level of down regulation of YBL111C.

Download:

Fig 3. The regulatory clustering heatmap related to genes targeted by identified transcription factors Hap4p and Tup1 and TOS8.

The cluster is represented as the log mRNA ratio of each target gene in each regulator mutant.

https://doi.org/10.1371/journal.pone.0259476.g003

Validation of top-ranked genes through different S. cerevisiae strains datasets

The top-ranked genes derived by supervised attribute weighting models showed high reproducibility in an independent microarray experiment related to evolved ethanol-tolerant clones derived from S. cerevisiae strains CEN.PK with high ethanol production rate, in comparison with the reference strain. Similar results were obtained from the evaluation of informative identified genes based on GSE5185 dataset which belongs to higher ethanol producing mutant strain spt15 in comparison with its reference. Noticeably, the captured variance with the first two components on 171 probe sets was as high as 78.60% and 70.40% in the GSE5185 and GSE78759 datasets, reinforcing that the identified top-ranked genes were able to accurately discriminate between low ethanol samples and high ethanol ones (S1 Fig). Also, according to LOOCV analysis result, the top-ranked genes were able to identify low and high ethanol samples with classification accuracy of 100%. The classification performance was further evaluated based on the following parameters: Precision = 1; Recall = 1; F-Measure = 1; and ROC Area = 1 (S1G File).

Discussion

In this study, machine learning and decision tree models were used to analyze the transcriptome of S. cerevisiae during the fermentation process in two conditions: repressed ethanol production and high ethanol production supplemented with Cu²⁺ and Mg²⁺ respectively. Indeed, for the most accurate prediction methods, we used both supervised and unsupervised models. In summary, we used 11 supervised models to achieve high accuracy results. In addition, a PCA analysis as an unsupervised model and a hierarchical clustering heat map were used to validate the 171 top-ranked probe sets identified by supervised-based models. In parallel, to validate the reliability of identified probe sets in another S. cerevisiae strain, we used to evaluate the reproducibility of the results by exploring the discriminative potential of them via PCA and LOOCV analysis based on different gene expression datasets related to another S. cerevisiae strain and different experiment. LOOCV and unsupervised analysis of the data using PCA demonstrated the potential of 171 probe sets to discriminate between the low and high ethanol samples from independent microarray experiments. The captured variance with the first two components by PCA analysis on 171 probe sets was high (78.60% and 70.40%). These findings imply that AWAs applied to FCdb can substantially reduce the dimensionality of the data by eliminating probe sets that are unlikely to be linked to ethanol production, leading to identification of the most reliable probe sets associated with ethanol production.

Furthermore, we used pathway enrichment, transcription factor and regulatory analysis to validate the machine learning analysis results (Fig 4). According to RapidMiner-assisted analysis, some probe sets were identified as playing a distinct role in ethanol production. Nonetheless, it should be noted that the function of some identified probe sets has not yet been clarified, despite the fact that they may be critical in ethanol production. ADH5 or Alcohol dehydrogenase, which was weighted by 9 algorithms and classified in Glycolysis / Gluconeogenesis by KEGG enrichment analysis, contributes to ethanol production by reducing acetaldehyde to ethanol [38]. OLI1 was distinguished by ten algorithms and enriched by Oxidative phosphorylation term, which encodes F0-ATP synthase subunit c and generates ATP in yeast mitochondria [39]. Cu²⁺ in toxic concentration, is known to have a negative effect on mitochondrial respiratory components, as it repressed the respiratory chain in PC12 and liver cells at toxic doses [40, 41]. That is most likely the main reason for the down regulation of OLI1, which is an important component of the oxidative phosphorylation pathway when exposed to Cu²⁺ treatment. RNA-seq analysis revealed that this gene was enriched as a significant gene between the wild and high glucose tolerant mutant strains of S. cerevisiae [42]. In addition to this gene, COX1 has an AWA weight of 8 and is involved in the final electron chain reaction in the respiratory system [43]. It encodes one of the cytochrome c oxidase subunits and, like OLI1, has been shown to be repressed by Cu²⁺ treatment. PDAI encodes alpha subunit of pyruvate dehydrogenase and converts the pyruvate to acetyl-CoA through oxidative decarboxylation [44]. This gene was found to be enriched by the Glycolysis/Gluconeogenesis pathway by seven weighting algorithms used in this study. PDAI directs the pyruvate metabolism to Acetyl- COA in mitochondria to provide the TCA cycle substrate. In other words, directing the pyruvate to TCA cycle PDAI keeps pyruvate from being consumed in the fermentation process or ethanol production. PDAI was down regulated in Mg-containing medium, which accounts for improved ethanol production, and was upregulated in the repressed fermentation condition, by Cu. QCR6 is a subunit of cytochrome bc1 complex and contributes to oxidative phosphorylation. Cytochrome C is known to be activated by Cu metal ion [45]. QCR6 was up regulated, as expected, by Cu supplementation. Similarly, in Pichia stipites, cytochrome bc1 disruption resulted in increased ethanol production. [46]. Granados-Arvizu et al [47] also concluded that cytochrome bc₁ complex repression would be a promising way to enhance ethanol production in Saccharomyces stipites. ALD6 or Aldehyde dehydrogenases is activated by Mg and have a distinct role in the formation of acetate from pyruvate in an alternate pyruvate dehydrogenase bypass pathway [48]. Here, ALD6 expression was found to be increased with Mg supplementation, which corresponded to the activation of this enzyme by Mg²⁺. Since it consumes the acetaldehyde source that ADH enzymes can use to produce ethanol, deleting ALD6 via clustered regularly interspaced short palindromic repeats (CRISPR/Cas9) genome editing resulted in increased ethanol production [49].

Download:

Fig 4. The schematic illustrates the methodology of the study with summarized results.

https://doi.org/10.1371/journal.pone.0259476.g004

Based on decision tree analysis, 17 identified roots exhibited 100% performance, some of which have unknown molecular functions and have yet to be characterized. Surprisingly, OLI1 and CYC3 were identified by the highest attribute weighting algorithms (10 and 9), were enriched by the second most significant biochemical pathway, and were also identified as decision tree model roots with 100% performance. COX1 is also shown as a complete root, but it is identified using 8 weighting algorithms. As previously stated, OLI1 is an F0-ATP synthase subunit c that contributes to the electron transport chain. CYC3 is also known as Cytochrome c heme lyase and has a strong sensitivity to ethanol. Indeed, the null mutants for this gene showed ethanol sensitivity. Both of these genes are involved in ATP generation and are up regulated in Mg supplemented medium according to the results. Nonetheless, it has been established that Mg²⁺ has an effect on energy metabolism and ATP production in the cell [50].

The Kegg enrichment analysis demonstrated that the topmost significant pathways, porphyrin biosynthesis and oxidative phosphorylation, were enriched in CYC3, OLI1 and COX1. Interestingly, these genes were also identified based on AWAs and decision tree models analysis. Besides, Cell cycle and division, as well as Ribosome biogenesis, are identified as significant terms in the KEGG pathway enrichment. They may have an impact on ethanol production even though they do not directly contribute to the fermentation bioprocess. For example, in addition to its role in yeast cell growth and proliferation, which affects ethanol production, ribosome biogenesis is predicted to be associated with fermentation, and some related genes, such as SFP1 are thought to be involved in glycolysis control as well [38, 51]. Nonetheless, significant phosphatidylinositol signaling and mitogen-activated protein kinase (MAPK) signaling pathways identified in this study by enrichment analysis were reported to be responsible for cell proliferation/growth regulation and critical for stress responses [52, 53]. PKC1 which was attributed by 8 algorithms and remarkably enriched in phosphatidylinositol signaling system is a serine/threonine kinase which is suggested to have role in response to copper toxicity since it was upregulated in Cu²⁺ supplementation or reduced ethanol production according to heatmap clustering. Confirming this finding, Zhou et al. [52] reported that 5-hydroxymethyl-2-furaldehyde, which is toxic to industrial fermentative S. cerevisiae strain, increases the expression of PKC1 gene. Furthermore, according to AWAs analysis, some genes are involved in stress responses, cell growth and proliferation, protein synthesis, fatty acids and lipid metabolism, all of which may contribute to ethanol production efficiency. MRP8 was assigned by ten algorithms as a response to cell wall stress, and its expression has been reported to be induced under stress conditions [54]. Its function, however, is unknown. Here, Cu²⁺ supplementation also induces the expression of this gene in response to the stress condition caused by copper. GTR2, a GTPase subunit, was weighted using nine algorithms. It is suggested in this study that it contributes to tolerance response to Cu supplementation because it was up regulated by copper. As an implication for this result, the null mutant related to GTR gene showed decreased resistance to Zn metal at inhibitory amount [55].

According to the crucial role of TFs in gene expression regulation and to confirm the results obtained from attribute weighting algorithms analysis, the TFs and their targets were explored among 171 probe sets. According to the regulatory clustering analysis, Tup1 has a significant effect on the top-ranked target genes. Tup1 is a transcriptional repressor in S. cerevisiae has the ability to repress target genes via various molecular mechanisms, and it contributes to carbon catabolite repression of transcription by glucose [56, 57]. Regarding the results of this study on regulatory clustering analysis, the Tup1 mutant caused decreased expression in some of the target genes and up regulation in others. In details, according to the regulatory analysis the Tup1 deletion most significantly resulted in the downregulation of YBL111C whose biological function is not known. Although the Tup1 deletion affect the ADH5, the expression change was not significant. On the other hand, the TUP1 knock out resulted in significant upregulation of YAL039C (CYC3). Indeed, the CYC3 gene, which was confirmed by the greatest number of AWAs and a decision tree model, was also shown to be a top target of the transcription factor involved in ethanol production responses in this study. HAP4 is a transcription factor involved in the regulation of the respiratory genes’ expression and ethanol tolerance. The role of TUP1 and HAP4 in glucose fermentation have been studied and recently confirmed in thermos-tolerant yeast, Ogataea polymorpha [57]. Moreover, the overexpression of HAP4 gene caused enhanced glucose consumption and ethanol production in S. cerevisiae [58, 59]. In this study, the HAP4 gene was also identified as top-ranked gene attributed by nine AWAs. Although the results confirm its involvement in the identified probe sets regulation, it does not demonstrate significant up or down-regulation effect on the target genes. Altogether, OLI1, CYC3, COX1 and ADH5 were ranked as the most critical genes in the differentiation of two improved and repressed ethanol production conditions because they were the most frequently identified genes across analyses. These important findings shed light on the complex pathways and regulatory responses that genes use to contribute to ethanol production. However, additional experimental analysis could fully clarify the results. Overall, the findings of this study could be used to further investigate the possibility of improving ethanol through overexpression or knock out strategies. Further experimental investigations, including overexpression and knockout studies as well as gene expression analysis by qRT-PCR are required to confirm the identified genes.

Conclusion

This investigation provides a significant understanding of the S. cerevisiae cell molecular response under low and high ethanol production. Preforming eleven machine learning algorithms via RapidMiner on 6300 probe sets related to S. cerevisiae transcriptomic data under low and high ethanol production, 171 discriminative probe sets were identified. Besides, PCA and hierarchical clustering confirm the accuracy of the supervised discriminating methods. Through different computational analyses including attribute weighting algorithms, decision tree models, unsupervised models, pathway enrichment, and regulatory analysis, prominent genes such as OLI1, CYC3, COX1, and ADH5 were recognized to be involved in ethanol production level. The results of this study also provide insight into the potential of genes that could be utilized in ethanol production enhancement programs. However, further experimental evaluations are crucial to confirm the findings.

Supporting information

S1 File.

A and B. The list of the probe sets along with the attribute weighting algorithms (AWAs) through which the probe sets were identified. C and D. The decision tree models (performance and roots). E. The list of identified transcription factors and their targets. G. Leave-one-out cross-validation (LOOCV) analysis on the top-ranked discriminative genes.

https://doi.org/10.1371/journal.pone.0259476.s001

(XLSX)

S1 Fig. Two-dimensional plot related to the first two principal components based on 171 identified probe sets.

https://doi.org/10.1371/journal.pone.0259476.s002

(TIF)

Acknowledgments

We thank the Dr. Leyla Nazari, Crop and Horticultural Science Research Department, Fars Agricultural and Natural Resources Research and Education Center, for assistance in validating analysis of top- ranked genes (LOOCV analysis).

References

1. Moysés DN, Reis VCB, Almeida JRMD, Moraes LMPD. Torres FAG Xylose fermentation by Saccharomyces cerevisiae: challenges and prospects. Int. J. Mol. Sci. 2016; 17, 207. pmid:26927067
- View Article
- PubMed/NCBI
- Google Scholar
2. da Silva Filho EA, de Melo HF, Antunes DF, dos Santos SK, do Monte Resende A, Simoes DA, et al. Isolation by genetic and physiological characteristics of a fuel-ethanol fermentative Saccharomyces cerevisiae strain with potential for genetic manipulation. J Ind Microbiol Biotechnol. 2005; 32: 481–486. pmid:16175407
- View Article
- PubMed/NCBI
- Google Scholar
3. Louis E. Saccharomyces cerevisiae: gene annotation and genome variability, state of the art through comparative genomics. Methods Mol. Biol. 2011;759: 31–40. pmid:21863479
- View Article
- PubMed/NCBI
- Google Scholar
4. Proux-Wéra E, Armisén D, Byrne KP, Wolfe KH. A pipeline for automated annotation of yeast genome sequences by a conserved-synteny approach. BMC bioinform. 2012; 13, 237. pmid:22984983
- View Article
- PubMed/NCBI
- Google Scholar
5. Gohil N, Panchasara H, Patel S, Ramírez-García R, Singh V. Book review: recent advances in yeast metabolic engineering. Front. Bioeng. Biotechnol. 2017; 5, 71. https://doi.org/10.3389/fbioe.2017.00071.
- View Article
- Google Scholar
6. Kobayashi Y, Sahara T, Ohgiya S, Kamagata Y, Fujimori KE. Systematic optimization of gene expression of pentose phosphate pathway enhances ethanol production from a glucose/xylose mixed medium in a recombinant Saccharomyces cerevisiae. AMB Express. 2018; 8, 139. pmid:30151682
- View Article
- PubMed/NCBI
- Google Scholar
7. Liu K, Yuan X, Liang L, Fang J, Chen Y, He W, et al. Using CRISPR/Cas9 for multiplex genome engineering to optimize the ethanol metabolic pathway in Saccharomyces cerevisiae. Biochem. Eng. J. 2019;145: 120–126a. https://doi.org/10.1016/j.bej.2019.02.017.
- View Article
- Google Scholar
8. Ye S, Jeong D, Shon JC, Liu KH, Kim KH, Shin M, et al. Deletion of PHO13 improves aerobic l-arabinose fermentation in engineered Saccharomyces cerevisiae. J. Ind. Microbiol. Biotechnol. 2019;46: 1725–1731. pmid:31501960
- View Article
- PubMed/NCBI
- Google Scholar
9. de Souza RB, Silva RK, Ferreira DS, Junior SDSLP, de Barros Pita W, de Morais Junior MA. Magnesium ions in yeast: setting free the metabolism from glucose catabolite repression. Metallomics. 2016;8: 1193–1203. pmid:27714092
- View Article
- PubMed/NCBI
- Google Scholar
10. Lewis J.A, Elkon I.M, McGee M.A, Higbee A.J, Gasch A.P. Exploiting natural variation in Saccharomyces cerevisiae to identify genes for increased ethanol resistance. Genetics. 2010; 186: 1197–1205. pmid:20855568
- View Article
- PubMed/NCBI
- Google Scholar
11. Pejin JD, Mojović LV, Pejin DJ, Kocić-Tanackov SD, Savić DS, Nikolić SB, et al. Bioethanol production from triticale by simultaneous saccharification and fermentation with magnesium or calcium ions addition. Fuel. 2015; 142: 58–64. https://doi.org/10.1016/j.fuel.2014.10.077.
- View Article
- Google Scholar
12. Taloria D, Samanta S, Das S, Pututunda C. Increase in bioethanol production by random UV mutagenesis of S. cerevisiae and by addition of zinc ions in the alcohol production media. APCBEE Procedia. 2012;2: 43–49. https://doi.org/10.1016/j.apcbee.2012.06.009.
- View Article
- Google Scholar
13. Ko JK, Um Y, Lee SM. Effect of manganese ions on ethanol fermentation by xylose isomerase expressing Saccharomyces cerevisiae under acetic acid stress. Bioresour. Technol. 2016; 222: 422–430. pmid:27744166
- View Article
- PubMed/NCBI
- Google Scholar
14. Wang FQ, Gao C.J, Yang CY, Xu P. Optimization of an ethanol production medium in very high gravity fermentation. Biotechnol. Lett. 2007;29: 233–236. pmid:17091376
- View Article
- PubMed/NCBI
- Google Scholar
15. Zhao XQ, Bai Fw. Zinc and yeast stress tolerance: Micronutrient plays a big role. J. Biotechnol. 2012; 158: 176–183. pmid:21763361
- View Article
- PubMed/NCBI
- Google Scholar
16. Ku I, Ku S, Sakamoto T, Hasunuma T, Zhao X, Kondo A. Zinc, magnesium, and calcium ion supplementation confers tolerance to acetic acid stress in industrial Saccharomyces cerevisiae utilizing xylose. Biotechnol. J. 2014;12: 1519–1525. https://doi.org/10.1002/biot.201300553.
- View Article
- Google Scholar
17. Shi H, Jiang Y, Yang Y, Peng Y, Li C. Copper metabolism in Saccharomyces cerevisiae: An update. Biometals, 2020; pp.1–12. pmid:31970541
- View Article
- PubMed/NCBI
- Google Scholar
18. Farrés M, Piña B, Tauler R. LC-MS based metabolomics and chemometrics study of the toxic effects of copper on Saccharomyces cerevisiae. Metallomics. 2016; 8: 790–798. pmid:27302082
- View Article
- PubMed/NCBI
- Google Scholar
19. Lihua W, Wang R, Zhan J, Huang W. High levels of copper retard the growth of Saccharomyces cerevisiae by altering cellular morphology and reducing its potential for ethanolic fermentation. Int. J. Food Sci. Technol. 2020; 2720–2731. https://doi.org/10.1111/ijfs.14903.
- View Article
- Google Scholar
20. Teixeira DAR. A Computational Platform for Gene Expression Analysis. M.Sc. Thesis] Informatics and Computing Engineering (MIEC), FEUP; 2014. https://hdl.handle.net/10216/75202.
21. Abusamra H. A comparative study of feature selection and classification methods for gene expression data of glioma. Procedia Comput. Sci. 2013;23 5–14. https://doi.org/10.1016/j.procs.2013.10.003.
- View Article
- Google Scholar
22. Zinati Z, Alemzadeh A, KayvanJoo AH. Computational approaches for classification and prediction of P-type ATPase substrate specificity in Arabidopsis. Physiol. Mol. Biol. Plants. 2016;22: 163–174. pmid:27186030
- View Article
- PubMed/NCBI
- Google Scholar
23. Jack C, Liu H, Lin W, Tsai F. Identification of contributing genes of Huntington’s disease by machine learning. BMC Medical Genom. 2020;13: 1–11. https://doi.org/10.1186/s12920-020-00822-w.
- View Article
- Google Scholar
24. Huang T, Zhang J, Xu Z.P, Hu L.L, Chen L, Shao J.L, et al. Deciphering the effects of gene deletion on yeast longevity using network and machine learning approaches. Biochimie. 2012; 94:1017–1025. pmid:22239951
- View Article
- PubMed/NCBI
- Google Scholar
25. RapidMiner Documentation. https://docs.rapidminer.com/. Accessed 10 Sep 2017.
26. Cho S, Won H. Machine learning in DNA microarray analysis for cancer classification, In Proceedings of the First Asia-Pacific Bioinformatics Conference on Bioinformatics. 2003. 189–198.
27. Xing EP, Jordan MI, Karp RM. Feature selection for high-dimensional genomic microarray data. In: 18th Int’l Conference Machine Learning. 2001. 601–608.
28. Alagukumar S, Lawrance R. A selective analysis of microarray data using association rule mining. Procedia Comput. Sci, 2015;47: 3–12. https://doi.org/10.1016/j.procs.2015.03.177.
- View Article
- Google Scholar
29. Metsalu T, Vilo J. ClustVis: a web tool for visualizing clustering of multivariate data using Principal Component Analysis and heatmap. Nucleic Acids Res. 2015;43: W566–70. pmid:25969447
- View Article
- PubMed/NCBI
- Google Scholar
30. Chen EY, Tan CM, Kou Y, Duan Q, Wang Z, Meirelles GV, et al. Enrichr: interactive and collaborative HTML5 gene list enrichment analysis tool. BMC Bioinformatics. 2013;14:128. pmid:23586463
- View Article
- PubMed/NCBI
- Google Scholar
31. Kuleshov MV, Jones MR, Rouillard AD, Fernandez NF, Duan Q, Wang Z, et al. Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic Acids Res. 2019;44: w90–w97. https://doi.org/10.1093/nar/gkw377.
- View Article
- Google Scholar
32. Monteiro PT, Oliveira J, Pais P, Antunes M, Palma M, Cavalheiro M, et al. YEASTRACT+: a portal for cross-species comparative genomics of transcription regulation in yeasts. Nucleic Acids Res. 2020;48: D642–D649. pmid:31586406
- View Article
- PubMed/NCBI
- Google Scholar
33. Hu Z, Killion PJ, Iyer VR. Genetic reconstruction of a functional transcriptional regulatory network. Nat. Genet. 2007;39: 683–687. pmid:17417638
- View Article
- PubMed/NCBI
- Google Scholar
34. Reimand J, Vaquerizas JM, Todd A.E, Vilo J, Luscombe N.M. Comprehensive reanalysis of transcription factor knockout expression data in Saccharomyces cerevisiae reveals many new targets. Nucleic Acids Res. 2010; 38: 4768–4777. pmid:20385592
- View Article
- PubMed/NCBI
- Google Scholar
35. Turanlı-Yıldız B, Benbadis L, Alkım C, Sezgin T, Akşit A, Gökçe A, et al. In vivo evolutionary engineering for ethanol-tolerance of Saccharomyces cerevisiae haploid cells triggers diploidization. J Biosci Bioeng. 2017;124:309–318. pmid:28552194
- View Article
- PubMed/NCBI
- Google Scholar
36. Alper H, Moxley J, Nevoigt E, Fink GR, Stephanopoulos G. Engineering yeast transcription machinery for improved ethanol tolerance and production, Science. 2006; 314:1565–1568. pmid:17158319
- View Article
- PubMed/NCBI
- Google Scholar
37. Qi K, Xia X.X, Zhong J.J. Enhanced anti-oxidative activity and lignocellulosic ethanol production by biotin addition to medium in Pichia guilliermondii fermentation, Bioresour. Technol. 2015;189:36–43. pmid:25864029
- View Article
- PubMed/NCBI
- Google Scholar
38. Davydenko S, Meledina T, Mittenberg A, Shabelnikov S, Vonsky M, Morozov A. Proteomics answers which yeast genes are specific for baking, brewing, and ethanol production. Bioeng. 2020; 7: 147. https://doi.org/10.3390/bioengineering7040147.
- View Article
- Google Scholar
39. Somlo M, Reid RA, Krupa M. An oligomycin-resistant adenosine triphosphatase and its effects on cellular growth, mitochondrial oxidative phosphorylation and respiratory proton translocation in Saccharomyces cerevisiae. Biochem j. 1997; 162: 51–59. https://doi.org/10.1042/bj1620051.
- View Article
- Google Scholar
40. Su R, Wang R, Guo S, Cao H, Pan J, Li C, et al. In vitro effect of copper chloride exposure on reactive oxygen species generation and respiratory chain complex activities of mitochondria isolated from broiler liver. Biol. trace elem. Res. 2011; 144 668–677. pmid:21455704
- View Article
- PubMed/NCBI
- Google Scholar
41. Belyaeva EA, Sokolova TV, Emelyanova LV, Zakharova IO. Mitochondrial electron transport chain in heavy metal-induced neurotoxicity: effects of cadmium, mercury, and copper. Sci. World J. 2012;136063. https://doi.org/10.1100/2012/136063.
- View Article
- Google Scholar
42. Chen Y, Lu Z, Chen D, Wei Y, Chen X, Huang J, et al. Transcriptomic analysis and driver mutant prioritization for differentially expressed genes from a Saccharomyces cerevisiae strain with high glucose tolerance generated by UV irradiation. RSC adv. 2017;7: 38784–38797. https://doi.org/10.1039/C7RA06146C.
- View Article
- Google Scholar
43. Vijayraghavan S, Kozmin SG, Strope PK, Skelly DA, Lin Z, Kennell J, et al. Mitochondrial genome variation affects multiple respiration and nonrespiration phenotypes in Saccharomyces cerevisiae. Genetics. 2019;211: 773–786. pmid:30498022
- View Article
- PubMed/NCBI
- Google Scholar
44. Steensma HY, Holterman L, Dekker I. van Sluis CA, Wenzel TJ. Molecular cloning of the gene for the E1 alpha subunit of the pyruvate dehydrogenase complex from Saccharomyces cerevisiae. Eur. J. Biochem. 1990;191: 769–74. pmid:2202601
- View Article
- PubMed/NCBI
- Google Scholar
45. Vest KE, Leary SC, Winge DR, Cobine PA. Copper import into the mitochondrial matrix in Saccharomyces cerevisiae is mediated by Pic2, a mitochondrial carrier family protein. J. Biol. Chem. 2013;288: 23884–23892. (2013). pmid:23846699
- View Article
- PubMed/NCBI
- Google Scholar
46. Shi NQ, Davis B, Sherman F, Cruz J, Jeffries TW. Disruption of the cytochrome c gene in xylose‐utilizing yeast Pichia stipitis leads to higher ethanol production. Yeast. 1999;15: 1021–1030. https://doi.org/10.1002/(SICI)1097-0061(199908)15:11<1021::AID-YEA429>3.0.CO;2-V. pmid:10455226
- View Article
- PubMed/NCBI
- Google Scholar
47. Granados-Arvizu JA, Madrigal-Perez LA, Canizal-García M, González-Hernández JC, García-Almendárez BE, Regalado-González C. Effect of cytochrome bc1 complex inhibition during fermentation and growth of Scheffersomyces stipitis using glucose, xylose or arabinose as carbon sources. FEMS yeast res. 2019;19. pmid:30500899
- View Article
- PubMed/NCBI
- Google Scholar
48. Meaden PG, Dickinson FM, Mifsud A, Tessier W, Westwater J, Bussey H, et al. The ALD6 gene of Saccharomyces cerevisiae encodes a cytosolic, Mg²⁺ activated acetaldehyde dehydrogenase. Yeast. 1997;13 1319–27. https://doi.org/10.1002/(SICI)1097-0061(199711)13:14<1319::AID-YEA183>3.0.CO;2-T. pmid:9392076
- View Article
- PubMed/NCBI
- Google Scholar
49. Papapetridis I, van Dijk M, Dobbe AP, Metz B, Pronk JT, van Maris AJ. Improving ethanol yield in acetate-reducing Saccharomyces cerevisiae by cofactor engineering of 6-phosphogluconate dehydrogenase and deletion of ALD6. Microb. Cell factories. 2016; 15: 67. pmid:27118055
- View Article
- PubMed/NCBI
- Google Scholar
50. Yamanaka R, Tabata S, Shindo Y, Hotta K, Suzuki K, Soga T, et al. Mitochondrial Mg ²⁺ homeostasis decides cellular energy metabolism and vulnerability to stress. Sci. rep. 2016; 6: 30027. pmid:27458051
- View Article
- PubMed/NCBI
- Google Scholar
51. Cipollina C, van den Brink J, Daran-Lapujade P, Pronk JT, Porro D, de Winde JH. Saccharomyces cerevisiae SFP1: at the crossroads of central metabolism and ribosome biogenesis. Microbiology. 2008;154: 1686–1699. pmid:18524923
- View Article
- PubMed/NCBI
- Google Scholar
52. Zhou Q, Liu ZL, Ning K, Wang A, Zeng X, Xu J. Genomic and transcriptome analyses reveal that MAPK-and phosphatidylinositol-signaling pathways mediate tolerance to 5-hydroxymethyl-2-furaldehyde for industrial yeast Saccharomyces cerevisiae. Sci. rep. 2014; 4: 6556. pmid:25296911
- View Article
- PubMed/NCBI
- Google Scholar
53. Adhikari H, Cullen PJ. Role of phosphatidylinositol phosphate signaling in the regulation of the filamentous-growth mitogen-activated protein kinase pathway. Eukaryot. cell. 2015; 14: 427–440. pmid:25724886
- View Article
- PubMed/NCBI
- Google Scholar
54. Boorsma A, de Nobel H, ter Riet B, Bargmann B, Brul S, Hellingwerf KJ, et al. Characterization of the transcriptional response to cell wall stress in Saccharomyces cerevisiae.Yeast. 2004;21: 413–427. pmid:15116342.
- View Article
- PubMed/NCBI
- Google Scholar
55. Dziedzic SA, Caplan AB. Identification of autophagy genes participating in zinc-induced necrotic cell death in Saccharomyces cerevisiae. Autophagy. 2011; 7: 490–500. pmid:21317551
- View Article
- PubMed/NCBI
- Google Scholar
56. Mennella TA, Klinkenberg LG Zitomer RS. Recruitment of Tup1-Ssn6 by yeast hypoxic genes and chromatin-independent exclusion of TATA binding protein. Eukaryot. cell. 2003; 2: 1288–1303. pmid:14665463
- View Article
- PubMed/NCBI
- Google Scholar
57. Kurylenko O, Ruchala J, Kruk B, Vasylyshyn R, Szczepaniak J, Dmytruk K, et al. The role of Mig1, Mig2, Tup1 and Hap4 transcription factors in regulation of xylose and glucose fermentation in the thermotolerant yeast Ogataea polymorpha. FEMS Yeast Res. 2021; 21. foab029. pmid:33983391
- View Article
- PubMed/NCBI
- Google Scholar
58. Williams FE, Trumbly RJ. Characterization of TUP1, a mediator of glucose repression in Saccharomyces cerevisiae. Mol. cell. biol. 1990;10: 6500–6511. pmid:2247069
- View Article
- PubMed/NCBI
- Google Scholar
59. Shi X, Zou Y, Chen Y, Ying H. Overexpression of THI4 and HAP4 improves glucose metabolism and ethanol production in Saccharomyces cerevisiae. Front. Microbiol. 2018;9: 1444. pmid:29997610
- View Article
- PubMed/NCBI
- Google Scholar

[ref1] 1. Moysés DN, Reis VCB, Almeida JRMD, Moraes LMPD. Torres FAG Xylose fermentation by Saccharomyces cerevisiae: challenges and prospects. Int. J. Mol. Sci. 2016; 17, 207. pmid:26927067
View Article
PubMed/NCBI
Google Scholar

[2] View Article

[3] PubMed/NCBI

[4] Google Scholar

[ref2] 2. da Silva Filho EA, de Melo HF, Antunes DF, dos Santos SK, do Monte Resende A, Simoes DA, et al. Isolation by genetic and physiological characteristics of a fuel-ethanol fermentative Saccharomyces cerevisiae strain with potential for genetic manipulation. J Ind Microbiol Biotechnol. 2005; 32: 481–486. pmid:16175407
View Article
PubMed/NCBI
Google Scholar

[6] View Article

[7] PubMed/NCBI

[8] Google Scholar

[ref3] 3. Louis E. Saccharomyces cerevisiae: gene annotation and genome variability, state of the art through comparative genomics. Methods Mol. Biol. 2011;759: 31–40. pmid:21863479
View Article
PubMed/NCBI
Google Scholar

[10] View Article

[11] PubMed/NCBI

[12] Google Scholar

[ref4] 4. Proux-Wéra E, Armisén D, Byrne KP, Wolfe KH. A pipeline for automated annotation of yeast genome sequences by a conserved-synteny approach. BMC bioinform. 2012; 13, 237. pmid:22984983
View Article
PubMed/NCBI
Google Scholar

[14] View Article

[15] PubMed/NCBI

[16] Google Scholar

[ref5] 5. Gohil N, Panchasara H, Patel S, Ramírez-García R, Singh V. Book review: recent advances in yeast metabolic engineering. Front. Bioeng. Biotechnol. 2017; 5, 71. https://doi.org/10.3389/fbioe.2017.00071.
View Article
Google Scholar

[18] View Article

[19] Google Scholar

[ref6] 6. Kobayashi Y, Sahara T, Ohgiya S, Kamagata Y, Fujimori KE. Systematic optimization of gene expression of pentose phosphate pathway enhances ethanol production from a glucose/xylose mixed medium in a recombinant Saccharomyces cerevisiae. AMB Express. 2018; 8, 139. pmid:30151682
View Article
PubMed/NCBI
Google Scholar

[21] View Article

[22] PubMed/NCBI

[23] Google Scholar

[ref7] 7. Liu K, Yuan X, Liang L, Fang J, Chen Y, He W, et al. Using CRISPR/Cas9 for multiplex genome engineering to optimize the ethanol metabolic pathway in Saccharomyces cerevisiae. Biochem. Eng. J. 2019;145: 120–126a. https://doi.org/10.1016/j.bej.2019.02.017.
View Article
Google Scholar

[25] View Article

[26] Google Scholar

[ref8] 8. Ye S, Jeong D, Shon JC, Liu KH, Kim KH, Shin M, et al. Deletion of PHO13 improves aerobic l-arabinose fermentation in engineered Saccharomyces cerevisiae. J. Ind. Microbiol. Biotechnol. 2019;46: 1725–1731. pmid:31501960
View Article
PubMed/NCBI
Google Scholar

[28] View Article

[29] PubMed/NCBI

[30] Google Scholar

[ref9] 9. de Souza RB, Silva RK, Ferreira DS, Junior SDSLP, de Barros Pita W, de Morais Junior MA. Magnesium ions in yeast: setting free the metabolism from glucose catabolite repression. Metallomics. 2016;8: 1193–1203. pmid:27714092
View Article
PubMed/NCBI
Google Scholar

[32] View Article

[33] PubMed/NCBI

[34] Google Scholar

[ref10] 10. Lewis J.A, Elkon I.M, McGee M.A, Higbee A.J, Gasch A.P. Exploiting natural variation in Saccharomyces cerevisiae to identify genes for increased ethanol resistance. Genetics. 2010; 186: 1197–1205. pmid:20855568
View Article
PubMed/NCBI
Google Scholar

[36] View Article

[37] PubMed/NCBI

[38] Google Scholar

[ref11] 11. Pejin JD, Mojović LV, Pejin DJ, Kocić-Tanackov SD, Savić DS, Nikolić SB, et al. Bioethanol production from triticale by simultaneous saccharification and fermentation with magnesium or calcium ions addition. Fuel. 2015; 142: 58–64. https://doi.org/10.1016/j.fuel.2014.10.077.
View Article
Google Scholar

[40] View Article

[41] Google Scholar

[ref12] 12. Taloria D, Samanta S, Das S, Pututunda C. Increase in bioethanol production by random UV mutagenesis of S. cerevisiae and by addition of zinc ions in the alcohol production media. APCBEE Procedia. 2012;2: 43–49. https://doi.org/10.1016/j.apcbee.2012.06.009.
View Article
Google Scholar

[43] View Article

[44] Google Scholar

[ref13] 13. Ko JK, Um Y, Lee SM. Effect of manganese ions on ethanol fermentation by xylose isomerase expressing Saccharomyces cerevisiae under acetic acid stress. Bioresour. Technol. 2016; 222: 422–430. pmid:27744166
View Article
PubMed/NCBI
Google Scholar

[46] View Article

[47] PubMed/NCBI

[48] Google Scholar

[ref14] 14. Wang FQ, Gao C.J, Yang CY, Xu P. Optimization of an ethanol production medium in very high gravity fermentation. Biotechnol. Lett. 2007;29: 233–236. pmid:17091376
View Article
PubMed/NCBI
Google Scholar

[50] View Article

[51] PubMed/NCBI

[52] Google Scholar

[ref15] 15. Zhao XQ, Bai Fw. Zinc and yeast stress tolerance: Micronutrient plays a big role. J. Biotechnol. 2012; 158: 176–183. pmid:21763361
View Article
PubMed/NCBI
Google Scholar

[54] View Article

[55] PubMed/NCBI

[56] Google Scholar

[ref16] 16. Ku I, Ku S, Sakamoto T, Hasunuma T, Zhao X, Kondo A. Zinc, magnesium, and calcium ion supplementation confers tolerance to acetic acid stress in industrial Saccharomyces cerevisiae utilizing xylose. Biotechnol. J. 2014;12: 1519–1525. https://doi.org/10.1002/biot.201300553.
View Article
Google Scholar

[58] View Article

[59] Google Scholar

[ref17] 17. Shi H, Jiang Y, Yang Y, Peng Y, Li C. Copper metabolism in Saccharomyces cerevisiae: An update. Biometals, 2020; pp.1–12. pmid:31970541
View Article
PubMed/NCBI
Google Scholar

[61] View Article

[62] PubMed/NCBI

[63] Google Scholar

[ref18] 18. Farrés M, Piña B, Tauler R. LC-MS based metabolomics and chemometrics study of the toxic effects of copper on Saccharomyces cerevisiae. Metallomics. 2016; 8: 790–798. pmid:27302082
View Article
PubMed/NCBI
Google Scholar

[65] View Article

[66] PubMed/NCBI

[67] Google Scholar

[ref19] 19. Lihua W, Wang R, Zhan J, Huang W. High levels of copper retard the growth of Saccharomyces cerevisiae by altering cellular morphology and reducing its potential for ethanolic fermentation. Int. J. Food Sci. Technol. 2020; 2720–2731. https://doi.org/10.1111/ijfs.14903.
View Article
Google Scholar

[69] View Article

[70] Google Scholar

[ref20] 20. Teixeira DAR. A Computational Platform for Gene Expression Analysis. M.Sc. Thesis] Informatics and Computing Engineering (MIEC), FEUP; 2014. https://hdl.handle.net/10216/75202.

[ref21] 21. Abusamra H. A comparative study of feature selection and classification methods for gene expression data of glioma. Procedia Comput. Sci. 2013;23 5–14. https://doi.org/10.1016/j.procs.2013.10.003.
View Article
Google Scholar

[73] View Article

[74] Google Scholar

[ref22] 22. Zinati Z, Alemzadeh A, KayvanJoo AH. Computational approaches for classification and prediction of P-type ATPase substrate specificity in Arabidopsis. Physiol. Mol. Biol. Plants. 2016;22: 163–174. pmid:27186030
View Article
PubMed/NCBI
Google Scholar

[76] View Article

[77] PubMed/NCBI

[78] Google Scholar

[ref23] 23. Jack C, Liu H, Lin W, Tsai F. Identification of contributing genes of Huntington’s disease by machine learning. BMC Medical Genom. 2020;13: 1–11. https://doi.org/10.1186/s12920-020-00822-w.
View Article
Google Scholar

[80] View Article

[81] Google Scholar

[ref24] 24. Huang T, Zhang J, Xu Z.P, Hu L.L, Chen L, Shao J.L, et al. Deciphering the effects of gene deletion on yeast longevity using network and machine learning approaches. Biochimie. 2012; 94:1017–1025. pmid:22239951
View Article
PubMed/NCBI
Google Scholar

[83] View Article

[84] PubMed/NCBI

[85] Google Scholar

[ref25] 25. RapidMiner Documentation. https://docs.rapidminer.com/. Accessed 10 Sep 2017.

[ref26] 26. Cho S, Won H. Machine learning in DNA microarray analysis for cancer classification, In Proceedings of the First Asia-Pacific Bioinformatics Conference on Bioinformatics. 2003. 189–198.

[ref27] 27. Xing EP, Jordan MI, Karp RM. Feature selection for high-dimensional genomic microarray data. In: 18th Int’l Conference Machine Learning. 2001. 601–608.

[ref28] 28. Alagukumar S, Lawrance R. A selective analysis of microarray data using association rule mining. Procedia Comput. Sci, 2015;47: 3–12. https://doi.org/10.1016/j.procs.2015.03.177.
View Article
Google Scholar

[90] View Article

[91] Google Scholar

[ref29] 29. Metsalu T, Vilo J. ClustVis: a web tool for visualizing clustering of multivariate data using Principal Component Analysis and heatmap. Nucleic Acids Res. 2015;43: W566–70. pmid:25969447
View Article
PubMed/NCBI
Google Scholar

[93] View Article

[94] PubMed/NCBI

[95] Google Scholar

[ref30] 30. Chen EY, Tan CM, Kou Y, Duan Q, Wang Z, Meirelles GV, et al. Enrichr: interactive and collaborative HTML5 gene list enrichment analysis tool. BMC Bioinformatics. 2013;14:128. pmid:23586463
View Article
PubMed/NCBI
Google Scholar

[97] View Article

[98] PubMed/NCBI

[99] Google Scholar

[ref31] 31. Kuleshov MV, Jones MR, Rouillard AD, Fernandez NF, Duan Q, Wang Z, et al. Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic Acids Res. 2019;44: w90–w97. https://doi.org/10.1093/nar/gkw377.
View Article
Google Scholar

[101] View Article

[102] Google Scholar

[ref32] 32. Monteiro PT, Oliveira J, Pais P, Antunes M, Palma M, Cavalheiro M, et al. YEASTRACT+: a portal for cross-species comparative genomics of transcription regulation in yeasts. Nucleic Acids Res. 2020;48: D642–D649. pmid:31586406
View Article
PubMed/NCBI
Google Scholar

[104] View Article

[105] PubMed/NCBI

[106] Google Scholar

[ref33] 33. Hu Z, Killion PJ, Iyer VR. Genetic reconstruction of a functional transcriptional regulatory network. Nat. Genet. 2007;39: 683–687. pmid:17417638
View Article
PubMed/NCBI
Google Scholar

[108] View Article

[109] PubMed/NCBI

[110] Google Scholar

[ref34] 34. Reimand J, Vaquerizas JM, Todd A.E, Vilo J, Luscombe N.M. Comprehensive reanalysis of transcription factor knockout expression data in Saccharomyces cerevisiae reveals many new targets. Nucleic Acids Res. 2010; 38: 4768–4777. pmid:20385592
View Article
PubMed/NCBI
Google Scholar

[112] View Article

[113] PubMed/NCBI

[114] Google Scholar

[ref35] 35. Turanlı-Yıldız B, Benbadis L, Alkım C, Sezgin T, Akşit A, Gökçe A, et al. In vivo evolutionary engineering for ethanol-tolerance of Saccharomyces cerevisiae haploid cells triggers diploidization. J Biosci Bioeng. 2017;124:309–318. pmid:28552194
View Article
PubMed/NCBI
Google Scholar

[116] View Article

[117] PubMed/NCBI

[118] Google Scholar

[ref36] 36. Alper H, Moxley J, Nevoigt E, Fink GR, Stephanopoulos G. Engineering yeast transcription machinery for improved ethanol tolerance and production, Science. 2006; 314:1565–1568. pmid:17158319
View Article
PubMed/NCBI
Google Scholar

[120] View Article

[121] PubMed/NCBI

[122] Google Scholar

[ref37] 37. Qi K, Xia X.X, Zhong J.J. Enhanced anti-oxidative activity and lignocellulosic ethanol production by biotin addition to medium in Pichia guilliermondii fermentation, Bioresour. Technol. 2015;189:36–43. pmid:25864029
View Article
PubMed/NCBI
Google Scholar

[124] View Article

[125] PubMed/NCBI

[126] Google Scholar

[ref38] 38. Davydenko S, Meledina T, Mittenberg A, Shabelnikov S, Vonsky M, Morozov A. Proteomics answers which yeast genes are specific for baking, brewing, and ethanol production. Bioeng. 2020; 7: 147. https://doi.org/10.3390/bioengineering7040147.
View Article
Google Scholar

[128] View Article

[129] Google Scholar

[ref39] 39. Somlo M, Reid RA, Krupa M. An oligomycin-resistant adenosine triphosphatase and its effects on cellular growth, mitochondrial oxidative phosphorylation and respiratory proton translocation in Saccharomyces cerevisiae. Biochem j. 1997; 162: 51–59. https://doi.org/10.1042/bj1620051.
View Article
Google Scholar

[131] View Article

[132] Google Scholar

[ref40] 40. Su R, Wang R, Guo S, Cao H, Pan J, Li C, et al. In vitro effect of copper chloride exposure on reactive oxygen species generation and respiratory chain complex activities of mitochondria isolated from broiler liver. Biol. trace elem. Res. 2011; 144 668–677. pmid:21455704
View Article
PubMed/NCBI
Google Scholar

[134] View Article

[135] PubMed/NCBI

[136] Google Scholar

[ref41] 41. Belyaeva EA, Sokolova TV, Emelyanova LV, Zakharova IO. Mitochondrial electron transport chain in heavy metal-induced neurotoxicity: effects of cadmium, mercury, and copper. Sci. World J. 2012;136063. https://doi.org/10.1100/2012/136063.
View Article
Google Scholar

[138] View Article

[139] Google Scholar

[ref42] 42. Chen Y, Lu Z, Chen D, Wei Y, Chen X, Huang J, et al. Transcriptomic analysis and driver mutant prioritization for differentially expressed genes from a Saccharomyces cerevisiae strain with high glucose tolerance generated by UV irradiation. RSC adv. 2017;7: 38784–38797. https://doi.org/10.1039/C7RA06146C.
View Article
Google Scholar

[141] View Article

[142] Google Scholar

[ref43] 43. Vijayraghavan S, Kozmin SG, Strope PK, Skelly DA, Lin Z, Kennell J, et al. Mitochondrial genome variation affects multiple respiration and nonrespiration phenotypes in Saccharomyces cerevisiae. Genetics. 2019;211: 773–786. pmid:30498022
View Article
PubMed/NCBI
Google Scholar

[144] View Article

[145] PubMed/NCBI

[146] Google Scholar

[ref44] 44. Steensma HY, Holterman L, Dekker I. van Sluis CA, Wenzel TJ. Molecular cloning of the gene for the E1 alpha subunit of the pyruvate dehydrogenase complex from Saccharomyces cerevisiae. Eur. J. Biochem. 1990;191: 769–74. pmid:2202601
View Article
PubMed/NCBI
Google Scholar

[148] View Article

[149] PubMed/NCBI

[150] Google Scholar

[ref45] 45. Vest KE, Leary SC, Winge DR, Cobine PA. Copper import into the mitochondrial matrix in Saccharomyces cerevisiae is mediated by Pic2, a mitochondrial carrier family protein. J. Biol. Chem. 2013;288: 23884–23892. (2013). pmid:23846699
View Article
PubMed/NCBI
Google Scholar

[152] View Article

[153] PubMed/NCBI

[154] Google Scholar

[ref46] 46. Shi NQ, Davis B, Sherman F, Cruz J, Jeffries TW. Disruption of the cytochrome c gene in xylose‐utilizing yeast Pichia stipitis leads to higher ethanol production. Yeast. 1999;15: 1021–1030. https://doi.org/10.1002/(SICI)1097-0061(199908)15:11<1021::AID-YEA429>3.0.CO;2-V. pmid:10455226
View Article
PubMed/NCBI
Google Scholar

[156] View Article

[157] PubMed/NCBI

[158] Google Scholar

[ref47] 47. Granados-Arvizu JA, Madrigal-Perez LA, Canizal-García M, González-Hernández JC, García-Almendárez BE, Regalado-González C. Effect of cytochrome bc1 complex inhibition during fermentation and growth of Scheffersomyces stipitis using glucose, xylose or arabinose as carbon sources. FEMS yeast res. 2019;19. pmid:30500899
View Article
PubMed/NCBI
Google Scholar

[160] View Article

[161] PubMed/NCBI

[162] Google Scholar

[ref48] 48. Meaden PG, Dickinson FM, Mifsud A, Tessier W, Westwater J, Bussey H, et al. The ALD6 gene of Saccharomyces cerevisiae encodes a cytosolic, Mg²⁺ activated acetaldehyde dehydrogenase. Yeast. 1997;13 1319–27. https://doi.org/10.1002/(SICI)1097-0061(199711)13:14<1319::AID-YEA183>3.0.CO;2-T. pmid:9392076
View Article
PubMed/NCBI
Google Scholar

[164] View Article

[165] PubMed/NCBI

[166] Google Scholar

[ref49] 49. Papapetridis I, van Dijk M, Dobbe AP, Metz B, Pronk JT, van Maris AJ. Improving ethanol yield in acetate-reducing Saccharomyces cerevisiae by cofactor engineering of 6-phosphogluconate dehydrogenase and deletion of ALD6. Microb. Cell factories. 2016; 15: 67. pmid:27118055
View Article
PubMed/NCBI
Google Scholar

[168] View Article

[169] PubMed/NCBI

[170] Google Scholar

[ref50] 50. Yamanaka R, Tabata S, Shindo Y, Hotta K, Suzuki K, Soga T, et al. Mitochondrial Mg ²⁺ homeostasis decides cellular energy metabolism and vulnerability to stress. Sci. rep. 2016; 6: 30027. pmid:27458051
View Article
PubMed/NCBI
Google Scholar

[172] View Article

[173] PubMed/NCBI

[174] Google Scholar

[ref51] 51. Cipollina C, van den Brink J, Daran-Lapujade P, Pronk JT, Porro D, de Winde JH. Saccharomyces cerevisiae SFP1: at the crossroads of central metabolism and ribosome biogenesis. Microbiology. 2008;154: 1686–1699. pmid:18524923
View Article
PubMed/NCBI
Google Scholar

[176] View Article

[177] PubMed/NCBI

[178] Google Scholar

[ref52] 52. Zhou Q, Liu ZL, Ning K, Wang A, Zeng X, Xu J. Genomic and transcriptome analyses reveal that MAPK-and phosphatidylinositol-signaling pathways mediate tolerance to 5-hydroxymethyl-2-furaldehyde for industrial yeast Saccharomyces cerevisiae. Sci. rep. 2014; 4: 6556. pmid:25296911
View Article
PubMed/NCBI
Google Scholar

[180] View Article

[181] PubMed/NCBI

[182] Google Scholar

[ref53] 53. Adhikari H, Cullen PJ. Role of phosphatidylinositol phosphate signaling in the regulation of the filamentous-growth mitogen-activated protein kinase pathway. Eukaryot. cell. 2015; 14: 427–440. pmid:25724886
View Article
PubMed/NCBI
Google Scholar

[184] View Article

[185] PubMed/NCBI

[186] Google Scholar

[ref54] 54. Boorsma A, de Nobel H, ter Riet B, Bargmann B, Brul S, Hellingwerf KJ, et al. Characterization of the transcriptional response to cell wall stress in Saccharomyces cerevisiae.Yeast. 2004;21: 413–427. pmid:15116342.
View Article
PubMed/NCBI
Google Scholar

[188] View Article

[189] PubMed/NCBI

[190] Google Scholar

[ref55] 55. Dziedzic SA, Caplan AB. Identification of autophagy genes participating in zinc-induced necrotic cell death in Saccharomyces cerevisiae. Autophagy. 2011; 7: 490–500. pmid:21317551
View Article
PubMed/NCBI
Google Scholar

[192] View Article

[193] PubMed/NCBI

[194] Google Scholar

[ref56] 56. Mennella TA, Klinkenberg LG Zitomer RS. Recruitment of Tup1-Ssn6 by yeast hypoxic genes and chromatin-independent exclusion of TATA binding protein. Eukaryot. cell. 2003; 2: 1288–1303. pmid:14665463
View Article
PubMed/NCBI
Google Scholar

[196] View Article

[197] PubMed/NCBI

[198] Google Scholar

[ref57] 57. Kurylenko O, Ruchala J, Kruk B, Vasylyshyn R, Szczepaniak J, Dmytruk K, et al. The role of Mig1, Mig2, Tup1 and Hap4 transcription factors in regulation of xylose and glucose fermentation in the thermotolerant yeast Ogataea polymorpha. FEMS Yeast Res. 2021; 21. foab029. pmid:33983391
View Article
PubMed/NCBI
Google Scholar

[200] View Article

[201] PubMed/NCBI

[202] Google Scholar

[ref58] 58. Williams FE, Trumbly RJ. Characterization of TUP1, a mediator of glucose repression in Saccharomyces cerevisiae. Mol. cell. biol. 1990;10: 6500–6511. pmid:2247069
View Article
PubMed/NCBI
Google Scholar

[204] View Article

[205] PubMed/NCBI

[206] Google Scholar

[ref59] 59. Shi X, Zou Y, Chen Y, Ying H. Overexpression of THI4 and HAP4 improves glucose metabolism and ethanol production in Saccharomyces cerevisiae. Front. Microbiol. 2018;9: 1444. pmid:29997610
View Article
PubMed/NCBI
Google Scholar

[208] View Article

[209] PubMed/NCBI

[210] Google Scholar

Figures

Abstract

Introduction

Materials and methods

Data collection

Data cleaning

Attribute weighting algorithms

Weight by information gain and information gain ratio

Weight by rule

Weight by deviation, weight by correlation and weight by chi squared statistic

Weight by Gini index and weight by uncertainty

Weight by relief

Weight by SVM

Weight by PCA

Decision tree models

Unsupervised analysis of the top ranked probe sets derived by supervised AWAs

Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis

Exploring for transcription factors among top-ranked genes and regulator cluster analysis

Independent validation of top-ranked genes using different microarray expression datasets

Results

Ranking probe sets by AWAs

Decision tree models

Unsupervised analysis

Pathway enrichment analysis of genes

Identification of transcription factors and their targets

Validation of top-ranked genes through different S. cerevisiae strains datasets

Discussion

Conclusion

Supporting information

S1 File.

S1 Fig. Two-dimensional plot related to the first two principal components based on 171 identified probe sets.

Acknowledgments

References