Figure 1.
Schematic overview of MethyC-seq protocol and Methy-Pipe workflow.
(A) The workflow of MethyC-seq library preparation and sequencing protocol. (B) The workflow and functional models of Methy-Pipe. DMRs: differentially methylated regions.
Figure 2.
Principle of bisulfite sequencing read alignment by BSAligner.
Firstly, the low-quality bases and sequenced adaptors at the 3′ ends of the reads are removed. The preprocessed reads are then mapped to C-to-T converted reference genomes whose Burrows-Wheeler Transform (BWT) indices are created and loaded to computing memory before executing alignment. Paired-end reads and single-end reads use different alignment strategies: (1) For single-end reads, they are mapped to reference genome by allowing at most 2 mismatches and only uniquely mapped reads are kept for further analysis; (2) For the paired-end reads, in addition to considering the number of mismatches and aligned hits, the insert size between the paired-end reads are also taken into account (e.g., from 50 to 600 bases); (3) The ambiguous reads that are mapped to both Watson and Crick strands are removed. Finally, the alignments are outputted in a text file which records the aligned chromosomes, positions, mismatches as well as sequencing qualities etc.
Figure 3.
Principle of DMR detection by BSAnalyzer.
(A) Firstly, starting from one end of the genome to search for a seed region (i.e., 500 bps) using a sliding window. (B) If the seed region is located, Mann-Whitney test will be used to test if the seed region is a differentially methylated seed region. (C) Two adjacent differentially methylated seed regions are merged into one extended seed region (seed region extension). (D) Two discontinued differentially methylated regions are further merged together if they are within a certain distance (e.g. less than 1000 bps) for further differential methylation test using χ2 test.
Figure 4.
Performance evaluation of Methy-Pipe.
(A) A total of 1 million reads (75 bases) were randomly simulated with different rates of bisulfite conversion and aligned to human (GRCh37/hg19) genome. Methy-Pipe accurately detected various simulated methylation levels at a constant mapping efficiency which is not affected by the sequence context. The H (in CHG and CHH) denotes C, T, or A. (B) Density scatter plots are plotted to measure the correlation between the MDs from Methy-Pipe and the β value from Infinium Human Methylationa450 array for CpG loci that can be integrated by both sequencing and array platforms from Placenta.
Figure 5.
Summary of Methy-Pipe results from BSAnalyzer module.
(A) The plot of the base (A, C, G, T) frequency at each sequencing cycle. X-axis indicates the sequencing cycle. Y-axis indicates the base frequency. (B) The length distribution of the insert size of a paired-end bisulfite sequencing library. X-axis represents the insert size. Y-axis represents the percentage of insert with the indicated size. (C, D) Whole genome methylation profiling with fixed window approach for buffy coat sample (C) and placenta sample (D). Dots on the top are for the Watson strand and triangles on the bottom are for the Crick strand. (E, F) Whole genome methylation profiling within different sequence contexts. MDs at different sequence contexts, namely CAA, CAC, CAG, CAT, CCA, CCC, CCG, CCT, CGA, CGC, CGG, CGT, CTA, CTC, CTG, CTT, are calculated for buffy coat (E) and placenta (F), respectively. (G, H). The fractions of the methylated cytosines are calculated for 3 different sequence contexts for buffy coat (G) and placenta (H), respectively. Fractional methylated C is calculated as the proportion of the methylated cytosines at a particular sequence context over total methylated C sequenced. The results indicate that most of the methylated cytosines are from CpG dinucleotides, i.e. CGA, CGC, CGG and CGT. The H in CHG, CHH represents A, C, or T.
Figure 6.
Methylation profiling for different genomic regions.
BC: Buffy coat; PLN: Placenta.
Figure 7.
Performance evaluation of Methy-Pipe DMR identification algorithm using methylation data from Infinium HumanMethylation450 array.
(A) The proportion of hypermethylated regions identified by the array platform also detected by the Methy-Pipe. (B) The proportion of hypomethylated regions detected in array also identified by the Methy-Pipe. (C) Comparison of the methylation level unchanged regions between the array platform and Methy-Pipe. Hypo: hypomethylated regions. Hyper: hypermethylated regions. Unchanged: unchanged regions.
Figure 8.
Biological insights revealed by Methy-Pipe.
(A) A snapshot of methylation density in promoter and gene body regions of RASSF1A gene. A DMR in promoter region by comparing buffycoat and placenta tissue using Methy-Pipe is highlighted. (B) Genome-wide methylation profiling of the global methylation pattern for maternal buffy coat and the placenta samples. The range of MD shown is from 0% (innermost) to 100% (outermost) and the distance between two lines is 10%. (C) The distribution of DMRs identified by Methy-Pipe across different genomic features. (D) Gene ontology analysis of hypermethylated regions in placenta. Hypo: hypo-methylated. Hyper: hyper-methylated.