Table 1.
Next-generation RNA sequencing data from CHO cell lines analyzed.
Figure 1.
Workflow for the reference-based and the non-reference-based re-assembly of CHO cell transcripts.
The left side shows the reference-based pipeline, the right side the non-reference-based pipeline. Different colors represent the different processes: assembly steps, red; mapping steps, orange; filtering steps, light blue; clustering, yellow; raw data, light green; annotated transcriptome data, dark green. Final transcriptome data is publicly available via the GenDBE [26] and SAMS [27] web tools.
Figure 2.
Number of CHO cell transcripts assembled with Cufflinks, Trinity, and Oases.
K-mer sizes vary between 23 and 135 for the Oases assembly.
Figure 3.
Length distribution of the transcripts assembled with Cufflinks, Trinity, and Oases.
Figure 4.
Comparison of the proportions of correctly assembled transcripts and misassemblies.
All transcripts with significant BLASTp hit against the mouse reference protein set were classified into “correct” (red), “short” (green) and “false” (blue) assembled transcripts.
Figure 5.
“u80-metric” comparison of individual transcriptom assemblies.
The comparative u80-metric results for the single Cufflinks, Trinity and Oases assemblies, the combined assemblies, the results of the reference-based re-assembly (ref-based), the non-reference-based re-assembly (non-ref-based) and the final transcript set (final transcripts) are compared to two publicly available CHO cell transcript sets, Xu et al. [13] and Becker et al. [14].
Figure 6.
Unique u80-metric mouse proteins for the individual assemblies.
Almost all individual assemblies (52 of 59) have transcripts with an ungapped alignment covering a mouse protein by more than 80% that are not present in the other assemblies.
Table 2.
Estimation of the number of cluster with paralogous genes.
Figure 7.
Comparison of the different transcript sets.
The comparative metrics values of the combined assemblies, the results of the reference-based and non-reference-based pipelines and the final transcriptome set. For comparison the results of published CHO transcriptome sets are also shown (Xu et al proteins [13] and Becker et al de novo assembly [14]).
Figure 8.
Comparison of the public transcript set from Xu et(a) the reference-based re-assembly and (b) the non-reference-based re-assembly.
The Venn diagram (a) gives an overview of the predicted unique gene loci of both data sets. More than 1/3 of the transcripts are present in both sets. The venn diagram (b) shows the gene clusters created with the wcd [23] tool.
Figure 9.
GO category distribution of the gene clusters from the final transcript set using all annotated GO terms up to the second level.
Figure 10.
Screenshot of the GenDBE web-interface.
The upper main area of the web-interface shows a graphical representation of a genomic contig with the exon/intron structure of the annotated genes. Informations of a selected gene or transcript are shown in the lower part in the center. All available actions, that can be performed on the selected gene, are listed on the bottom right part of the main window. The menu on the top of the interface gives the user different means to browse the complete database.
Figure 11.
Screenshot of the SAMS web-interface showing a single transcript cluster.
The upper half of the interface shows a graphical representation of a possible splicing-graph of the cluster. The lower part lists all transcripts of the cluster with some information about the functional annotation of the transcript. Links to browse the database are on the left side of the interface.