HPG-DHunter: an ultrafast friendly tool for DMR detection and visualization

摘要

Bisulphite-treated DNA methylation analysis requires specific treatment of DNA that modifies its sequence, as well as software tools for its analysis. Bisulphite treatment converts unmethylated cytosines (Cs) into thymines (Ts), which gives rise to C-to-T changes in DNA sequence after sequencing, while leaving methylated cytosines (5mCs) unchanged. By aligning and comparing bisulphite sequencing reads to the genomic DNA sequence, it is possible to infer DNA methylation patterns at base pair-resolution ([1]). Hydroxymethylated samples can also be obtained from different methods, the Ten-eleven translocation (TET) Assisted Bisulfite Sequencing (TAB-Seq) ([2, 3]), which produces Ts in methylated and unmethylated Cs and maintains as C the hydroxymethylated Cs (5hmC), and the oxidative bisulphate sequencing (oxBs-seq) ([4]). Different software tools have been proposed for DNA methylation analysis like RRBSMAP ([5]), the widely extended tool Bismark ([6]), or the most recent tools HPG-Methyl ([7, 8]). These tools provide single-base information of the alignment and the methylation status of each input sequence (or read). However, all these software tools yield the results as text files which usually have sizes of tens of Gigabytes, and follow the Sequence Alignment/Map (SAM) format or the Binary Alignment Map (BAM) format. Other tools like HPG-HMapper ([9, 10]) uses the methylation information of each base for each read present in the BAM files to build a DNA methylation map which gives information about the methylation level for each base of the reference genome. This map is yielded as one csv file for each chromosome in the species. Biomedical researches then have to compare the methylation level information in these files at different scales (DNA segments, CpG islands or coding regions, DNA chromosomes, etc.) besides the base-pair resolution, comparing also the results coming from different samples. Thus, many different tools for processing, displaying the methylation results and discovering differentially methylated regions (DMRs) have been proposed ([11–16]). However, most of these tools are based on statistical techniques, adding a high computational workload when applied to huge BAM or SAM files. As a result, the execution time required by these tools is large, adding an excessive delay from the moment of the sample extraction from the DNA sequencer until the moment when significant information is provided to the doctor or biomedical researcher. Also, the visualization of these data is far from being interactive. Another disadvantage of these tools is that many of them are essentially R scripts, which requires programming skills from the user. Finally, as some comparative study shows [17], these methods show more accurate results when identifying simulated DM regions that are long and have small within-group variation, but they have low concordance, probably due to the different approaches they have used for DM identification. Thus, they yield very low concordance when used with real data.