The cell-free DNA (cfDNA) methylation profile in liquid biopsy has been utilized to diagnose early-stage disease and estimate therapy response. However, typical clinical procedures are capable of purifying only very small amounts of cfDNA. Whole-genome bisulfite sequencing (WGBS) is the gold standard for measuring DNA methylation; however, WGBS using small amounts of fragmented DNA introduces a critical challenge for data analysis, namely a low-mapping ratio. The resulting low sequencing depth and low coverage of CpG sites genome-wide is a bottleneck for the clinical application of cfDNA-based WGBS assays. We developed LiBis (Low-input Bisulfite Sequencing), a novel method for low-input WGBS data alignment. By dynamically clipping initially unmapped reads and remapping clipped fragments, we judiciously rescued those reads and uniquely aligned them to the genome. By substantially increasing the mapping ratio by up to 88%, LiBis dramatically improved the number of informative CpGs and the precision in quantifying the methylation status of individual CpG sites. LiBis significantly improved the cost efficiency of low-input WGBS experiments by dynamically removing contamination introduced by random priming. The high sensitivity and cost effectiveness afforded by LiBis for low-input samples will allow the discovery of genetic and epigenetic features suitable for downstream analysis and biomarker identification using liquid biopsy.

BACKGROUND: Epigenetic regulation is essential in regulating gene expression across a variety of biological processes. Many high-throughput sequencing technologies have been widely used to generate epigenetic data, such as histone modification, transcription factor binding sites, DNA modifications, chromatin accessibility, and etc. A large scale of epigenetic data is stored in NCBI Gene Expression Omnibus (GEO). However, it is a great challenge to reanalyze these large scale and complex data, especially for researchers who do not specialize in bioinformatics skills or do not have access to expensive computational infrastructure. RESULTS: GsmPlot can simply accept GSM IDs to automatically download NCBI data or can accept user's private bigwig files as input to plot the concerned data on promoters, exons or any other user-defined genome locations and generate UCSC visualization tracks. By linking public data repository and private data, GsmPlot can spark data-driven ideas and hence promote the epigenetic research. CONCLUSIONS: GsmPlot web server allows convenient visualization and efficient exploration of any NCBI epigenetic data in any genomic region without need of any bioinformatics skills or special computing resources. GsmPlot is freely available at https://gsmplot.deqiangsun.org/.

MOTIVATION: Bisulfite sequencing (BS-seq) has emerged as the gold standard to study genome-wide DNA methylation at single-nucleotide resolution. Quality control (QC) is a critical step in the analysis pipeline to ensure that BS-seq data are of high quality and suitable for subsequent analysis. Although several QC tools are available for next-generation sequencing data, most of them were not designed to handle QC issues specific to BS-seq protocols. Therefore, there is a strong need for a dedicated QC tool to evaluate and remove potential technical biases in BS-seq experiments. RESULTS: We developed a package named BSeQC to comprehensively evaluate the quality of BS-seq experiments and automatically trim nucleotides with potential technical biases that may result in inaccurate methylation estimation. BSeQC takes standard SAM/BAM files as input and generates bias-free SAM/BAM files for downstream analysis. Evaluation based on real BS-seq data indicates that the use of the bias-free SAM/BAM file substantially improves the quantification of methylation level. AVAILABILITY AND IMPLEMENTATION: BSeQC is freely available at: http://code.google.com/p/bseqc/.

SUMMARY: Reduced representation bisulfite sequencing (RRBS) is a powerful yet cost-efficient method for studying DNA methylation on a genomic scale. RRBS involves restriction-enzyme digestion, bisulfite conversion and size selection, resulting in DNA sequencing data that require special bioinformatic handling. Here, we describe RRBSMAP, a short-read alignment tool that is designed for handling RRBS data in a user-friendly and scalable way. RRBSMAP uses wildcard alignment, and avoids the need for any preprocessing or post-processing steps. We benchmarked RRBSMAP against a well-validated MAQ-based pipeline for RRBS read alignment and observed similar accuracy but much improved runtime performance, easier handling and better scaling to large sample sets. In summary, RRBSMAP removes bioinformatic hurdles and reduces the computational burden of large-scale epigenome association studies performed with RRBS. AVAILABILITY: http://rrbsmap.computational-epigenetics.org/ http://code.google.com/p/bsmap/ CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.