- Open Access
VAliBS: a visual aligner for bisulfite sequences
© The Author(s). 2017
- Published: 16 October 2017
Methylation is a common modification of DNA. It has been a very important and hot topic to study the correlation between methylation and diseases in medical science. Because of the special process with bisulfite treatment, traditional mapping tools do not work well with such methylation experimental reads. Traditional aligners are not designed for mapping bisulfite-treated reads, where the un-methylated ‘C’s are converted to ‘T’s.
In this paper, we develop a reliable and visual tool, named VAliBS, for mapping bisulfate sequences to a genome reference. VAliBS works well even on large scale data or high noise data. By comparing with other state-of-the-art tools (BisMark, BSMAP, BS-Seeker2), VAliBS can improve the accuracy of bisulfite mapping. Moreover, VAliBS is a visual tool which makes its operations more easily and the alignment results are shown with colored marks which makes it easier to be read. VAliBS provides fast and accurate mapping of bisulfite-converted reads, and a friendly window system to visualize the detail of mapping of each read.
VAliBS works well on both simulated data and real data. It can be useful in DNA methylation research. VALiBS implements an X-Window user interface where the methylation positions are visual and the operations are friendly.
- DNA methylation
- Bisulfite mapping
- Visual alignment
Cytosine in CG dinucleotide (C in the 5′ end, G in the 3′ end) can be converted into 5-methyl cytosine under the enzyme by adding a methyl, which is called cytosine methylation of DNA. Cytosine methylation widely influences the expression of genes. Recent researches have shown that methylation is associated with many diseases, such as cancer, and methylation is heritable, which can be passed on to children from their parents . One popular method in cytosine methylation research is bisulfite treatment.
By comparing un-bisulfite-treated to bisulfite-treated sequences, we can identify where cytosine is methylated. It has been shown by Deng et al.  that targeted bisulfite sequencing reveals changes in DNA methylation associated with nuclear reprogramming. Bisulfite conversion of genomic DNA combined with next-generation sequencing has been widely used to measure the methylation state of a whole genome and the study of complex diseases, such as cancer. A survey for analyzing the cancer methylome through targeted bisulfite sequencing is reported in reference . Now the genome-wide bisulfite sequencing can also be used in single-cell , which provides a robust platform for molecular diagnotics . Gu et al. optimized bisulfite sequencing and analyzed clinical samples with genome-scale DNA methylation mapping at single-nucleotide resolution . Thus, it is of great interest to find the correct positions of bisulfite reads.
Recent years, great progresses have been made in the mapping tools for un-bisulfite-treated sequences . Several tools have been developed including Bowtie , Bowtie2 , BWA , RAUR , etc., which have been used widely in the genome assembly [12, 13], contig error correction  and structural variation detection . The existing mapping tools for bisulfite-treated sequences can be categorized into two groups: wild-card aligners and three-letter aligners [16, 17]. The common character of wild-card aligners is to replace cytosines in the sequenced reads with wild-card Y nucleotides to allow bisulfite mismatches. BSMAP , RMAPBS , GSNAP , and Segemehl  all employed this strategy. BSMAP was developed by Xi et al. based on a modified version of a general mapping tool SOAP . BSMAP  adopted hashing and fast lookup methods to the octamer seeds converted from the reference genome and used a bit-mapping strategy to highlight mismatches from methylation and sequencing errors. RMAPBS  was developed by Smith et al. based on the RMAP program for mapping single-end bisulphite reads. GSNAP  was developed by Wu et al., which can be used for both single- and paired-end reads mapping and can detect short- and long-distance splicing, including interchromosomal splicing.
On the other hand, three-letter aligners, such as bsmapper (https://sourceforge.net/projects/bsmapper/), BS-Seeker , Bismark , BRAT , BRAT-BW  and MethylCoder , convert C to T in both sequenced reads and genome reference prior to performing the reads mapping by using modified conventional aligners. Bismark  was developed by Krueger et al. based on the mapping tool Bowtie2 , which was not only for bisulfite sequence mapping but also for methylation call. Three-letter strategy makes it easier to reuse non-bisulfite aligner as an internal module, with these non-bisulfite aligners improved, it is convenient to replace the internal module. BRAT-BW  developed by Harris et al. is a fast, accurate and memory-efficient mapping tool which maps the bisulfite-treated short reads by using FM-index (Burrows-Wheeler transform). MethylCoder  developed by Pedersen et al. is a flexible software tool for mapping bisulfite-treated short reads, which supports both paired- and single-end reads in color space or nucleotide formats. MethylCoder provides the option to user with two existing short-read aligners: Bowtie  and GSNAP .
Most of the three-letter aligners are fast, accurate, memory-efficient, and flexible. They are based on the modified conventional aligners and have been widely used. So, we believe that new tools for bisulfite-treated sequences with higher recall and precision could be implemented with the development of general mapping tools. In this paper, we developed a new tool VAliBS based on the three-letter strategy for mapping bisulfite-treated short reads by integrating two latest excellent mapping tools of Bowtie2  and BWA . Moreover, VAliBS is a visual tool, in which the alignment results are shown with colored marks which make it easier to be read.
According to Fig. 1,we know that the sequenced reads are bisulfite treated, and the reference is un-bisulfite treated. In the case that maps the reads to references directly without any processing, converted base positions will be regarded as mismatches and result in large scale match failure. To avoid these cases, we employee the widely used three-letter strategy. Three-letter strategy will mask the difference between bisulfite converted and un-bisulfite converted bases. Specificly, it masks the difference between C and T artificially, which in the other strand is G and A. As a result, for every reference, we make two copies for it, one converting all C to T, the other one converting all G to A; for every read, we conduct the same process. Now we get double references and reads and could observe that the conversion takes some pseudo mapping. For example, because C and T have no difference in the mapping process, read AGACCCATG is mapped into AGATTTATG on reference by mistakes. However, according to the methylation process, there only exists C-to-T conversion, and does not exist T-to-C conversion. These issues can be addressed in the post-processing stage. In the pre-processing, a conversion operation was implemented both for the genome reference and for the sequencing reads. Since C turns into T in the original strands of bisulfite-treated reads and G turns into A on the new reverse complementary strands, we hence use two types of base conversions: one is converting C to T, and the other is converting G to A.
Overlap of mapping rate between Bowtie2 and BWA on Illumina reads
Illumina 75 bp
Illumina 100 bp
From the analysis results we can see that Bowtie2 works very well on low-noise data, but has a lower recall for high-noise data, and BWA employs a heuristic method and always returns a high recall both on the low and high-noise data. Thus, we first use Bowtie2 to get a very reliable mapping set and then use BWA to the un-mapping reads. On the other hand, tools like Bowtie2 and BWA execute bi-directional mapping by default. It means that they try to map the reverse and complementary strands of reads into the reference. After the three-letter conversion, we expect to have the direction of mapping, we just want to see read_c2t (reads only contain A,T,G) mapping into reference_c2t (reference also only contains A,T,G) forward, not except the read_c2t (reads contain A,C,G) also mapping into refernceence_c2t after reverse and complementary conversion, i.e., read_c2t will map into reference_c2t only if read_c2t and reference_c2t are in the same strand. Therefore, we should forbid the optional of automatic bi-directional mapping. Moreover, to ensure no possible mapped reads are missed, we try to keep more mappings even those of false mappings. Actually, these false mappings will be filtered in the post-processing.
In the post-processing, we also consider the mismatches with SNP tolerance by inputting SNP files to avoid filtering correct results. In addition, we need to merge the mapping results of Bowtie2 and BWA. Due to the introduction of conversion operation in VAliBS, it may generate multiple mapping results for the same original unconverted read. The repeated results will be removed.
In order to validate the effectiveness of VAliBS, we compare it with other popular bisulfite mapping tools: Bismark , BS-Seeker2 , and BSMAP . VAliBS, Bismark, and BS-Seeker2 are all the three-letter-based approaches. Bismark  is an efficient bisulfite mapping tool based on the modification of Bowtie2. BS-Seeker2  is an updated version of BS-Seeker, which further improves the mappability by using local alignment. BSMAP , on the contrast, is a method based on the wild-card approach. We compared them on both the simulation data and the real data.
The simulation data and real data are used as the same as in BSSeeker2 . Since our tool VAliBS for RRBS data did not have special treatment, we did not test RRBS data. Only WGBS data was used in our experiments. Two kinds of simulated sequences (error-free and error-containing) were used. For each kind of simulated sequences, both single-end and paired-end data were generated. The simulated error-containing sequences were converted with 1% failure, to which the sequencing errors by cycles were also added . The error-free simulated sequences were converted faithfully with no sequencing error. The single end of real data was from the published data sets, SRR299053 (mouse) and the paired-end of real data was from SRR306438 (human) .
Performance on simulation data
Comparison of VAliBS, Bismark, BS-Seeker2, and BSMAP on simulation data
The mappability (abbreviated as map in Table 2) is defined as the percentage of reads that are uniquely mapped over all reads. The correct mappability (abbreviated as c-map in Table 2) is defined as the percentage of corrected unique mapping.
VAliBS integrated Bowtie2 and BWA, which has greater flexibility and obtains different results with different parameters. As both Bismark and BS-Seeker2 used Bowtie2, we listed the results of VAliBS only by using Bowtie2. For comparison, the recommended parameters of Bowtie2 were used to evaluate the mappability and correct mappability of VAliBS, Bismark, and BS-Seeker2.
From Table 2 we can see that VAliBS, Bismark, BS-Seeker2, and BSMAP all work well on the single-end data for both error-free and error-containing data. Compared to the application on the simulated error-free data, the mappability and correct mappability of all the four bisulfite mapping tools slightly descend when being applied on the simulated data with noise. When being applied on the paired data, the mappability and correct mappability of VAliBS are much higher than those of Bismark, BS-Seeker2, and BSMAP.
Performance on real data
Comparison of VAliBS, Bismark, BS-Seeker2, and BSMAP on single-end data (SRR299053/mouse) and paired-end data (SRR306438/human)
Features supported by Bismark, BS-Seeker, BS-Seeker2, BSMAP, and VAliBS
Linux, Unix, Mac
Linux, Unix, Mac
SAM BAM Native
DNA methylation is very important to the research of diseases. In this paper, we have designed and implemented a visual tool VAliBS for bisulfite sequence alignment based on base conversions. VAliBS is fast, memory-efficient and reliable, which can be useful in DNA methylation research. More importantly, VAliBS is a visual tool where the alignment results and the methylation positions are visual while the operations are friendly. In addition, pre-processing and post-processing are decoupled with Bowtie2 and BWA, to make them easily updating modularity. As MapReduce frame has been used widely in bioinformatics , the efficiency performance of VAliBS can even be improved by parallel processing in the future.
VAliBS is based on the open source software BWA and Bowtie2. We would like to thank Dr. H. Li, and Dr. R. Durbinfor the source code and documentation of BWA and also are thankful to Dr. B. Langmead and coworkers for the source code and documentation of Bowtie2.
Part of this paper, an abridged two-page abstract, has been published in the Lecture notes in computer science: Bioinformatics research and applications .
This work was funded by the National Natural Science Foundation of China under Grants No. 61379108 and No.61232001. The National Natural Science Foundation of China supported the publication fee of this paper.
Availability of data and materials
About this supplement
This article has been published as part of BMC Bioinformatics Volume 18 Supplement 12, 2017: Selected articles from the 12th International Symposium on Bioinformatics Research and Applications (ISBRA-16): bioinformatics. The full contents of the supplement are available online at <https://bmcbioinformatics.biomedcentral.com/articles/supplements/volume-18-supplement-12>.
ML and XDY designed the schematic diagram of VAliBS including pre-processing, mapping, and post-processing. PH and XDY obtained the data and implemented the tool. ML and XDY analyzed the experimental results. ML, PH, XDY, JXW YP and FXW participated in revising the draft. All authors have read and approved the final manuscript.
Ethics approval and consent to participate
Consent for publication
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Li E, Beard C, Jaenisch R. Role for DNA methylation in genomic imprinting [J]. Nature. 1993;366(6453):362–5.View ArticlePubMedGoogle Scholar
- Deng J, Shoemaker R, Xie B, et al. Targeted bisulfite sequencing reveals changes in DNA methylation associated with nuclear reprogramming[J]. Nat Biotechnol. 2009;27(4):353–60.View ArticlePubMedPubMed CentralGoogle Scholar
- Lee EJ, Luo J, Wilson JM, et al. Analyzing the cancer methylome through targeted bisulfite sequencing[J]. Cancer Lett. 2013;340(2):171–8.View ArticlePubMedGoogle Scholar
- Smallwood SA, Lee HJ, Angermueller C, et al. Single-cell genome-wide bisulfite sequencing for assessing epigenetic heterogeneity[J]. Nat Methods. 2014;11(8):817–20.View ArticlePubMedPubMed CentralGoogle Scholar
- Gu H, Smith ZD, Bock C, et al. Preparation of reduced representation bisulfite sequencing libraries for genome-scale DNA methylation profiling[J]. Nat Protoc. 2011;6(4):468–81.View ArticlePubMedGoogle Scholar
- Gu H, Bock C, Mikkelsen TS, et al. Genome-scale DNA methylation mapping of clinical samples at single-nucleotide resolution[J]. Nat Methods. 2010;7(2):133–6.View ArticlePubMedPubMed CentralGoogle Scholar
- Zou Q, Hu Q, Guo M, Wang G. HAlign: Fast multiple similar DNA/RNA sequence alignment based on the centre star strategy[J]. Bioinformatics. 2015;31(15):2475–81.View ArticlePubMedGoogle Scholar
- Langmead B, Trapnell C, Pop M, et al. Ultrafast and memory-efficient alignment of short DNA sequences to the human genome[J]. Genome Biol. 2009;10(3):1–10.View ArticleGoogle Scholar
- Langmead B, Salzberg SL. Fast gapped-read alignment with Bowtie 2[J]. Nat Methods. 2012;9(4):357–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Li H, Durbin R. Fast and accurate long-read alignment with Burrows–Wheeler transform[J]. Bioinformatics. 2010;26(5):589–95.View ArticlePubMedPubMed CentralGoogle Scholar
- Peng X, Wang J, Zhang Z, et al. Re-alignment of the unmapped reads with base quality score[J]. Bmc Bioinformatics. 2014;16(Suppl 5):1–10.Google Scholar
- Luo J, Wang J, Zhang Z, et al. BOSS: a novel scaffolding algorithm based on optimized scaffold graph[J]. Bioinformatics. 2017;33(2):169–76.View ArticlePubMedGoogle Scholar
- Li M, Liao Z, He Y, et al. ISEA: iterative seed-extension algorithm for de novo assembly using paired-end information and insert size distribution[J]. IEEE/ACM Trans Comput Biol Bioinform. 10.1109/TCBB.2016.2550433.
- Li M, Wu B, Yan X, et al. PECC: correcting contigs based on paired-end read distribution. Comput Biol Chem. DOI: 10.1016/j.compbiolchem.15.03.012.
- Zhang Z, Wang J, Luo J, et al. Sprites: detection of deletions from sequencing data by re-aligning split reads[J]. Bioinformatics. 2016: btw053.Google Scholar
- Bock C. Analysing and interpreting DNA methylation data[J]. Nat Rev Genet. 2012;13(10):705–19.View ArticlePubMedGoogle Scholar
- Adusumalli S, Omar MFM, Soong R, et al. Methodological aspects of whole-genome bisulfite sequencing analysis[J]. Brief Bioinform. 2015;16(3):369–79.View ArticlePubMedGoogle Scholar
- Xi Y, Li W. BSMAP: whole genome bisulfite sequence MAPping program[J]. BMC Bioinformatics. 2009;10(1):1.View ArticleGoogle Scholar
- Smith AD, Chung WY, Hodges E, et al. Updates to the RMAP short-read mapping software[J]. Bioinformatics. 2009;25(21):2841–2.View ArticlePubMedPubMed CentralGoogle Scholar
- Wu TD, Nacu S. Fast and SNP-tolerant detection of complex variants and splicing in short reads[J]. Bioinformatics. 2010;26(7):873–81.View ArticlePubMedPubMed CentralGoogle Scholar
- Hoffmann S, Otto C, Kurtz S, et al. Fast mapping of short sequences with mismatches, insertions and deletions using index structures[J]. PLoS Comput Biol. 2009;5(9):e1000502.View ArticlePubMedPubMed CentralGoogle Scholar
- Li R, Li Y, Kristiansen K, et al. SOAP: short oligonucleotide alignment program[J]. Bioinformatics. 2008;24(5):713–4.View ArticlePubMedGoogle Scholar
- Harrison A, Parle-McDermott A. DNA methylation: a timeline of methods and applications[J]. Front Genet. 2011;2(74):1–13.Google Scholar
- Krueger F, Andrews SR. Bismark: a flexible aligner and methylation caller for Bisulfite-Seq applications[J]. Bioinformatics. 2011;27(11):1571–2.View ArticlePubMedPubMed CentralGoogle Scholar
- El-Maarri O. Methods: DNA methylation[M]//Peroxisomal disorders and regulation of genes. Springer US: 2003. p. 197-204.Google Scholar
- Harris EY, Ponts N, Le Roch KG, et al. BRAT-BW: efficient and accurate mapping of bisulfite-treated reads[J]. Bioinformatics. 2012;28(13):1795–6.View ArticlePubMedPubMed CentralGoogle Scholar
- Pedersen B, Hsieh TF, Ibarra C, et al. MethylCoder: software pipeline for bisulfite-treated sequences[J]. Bioinformatics. 2011;27(17):2435–6.View ArticlePubMedPubMed CentralGoogle Scholar
- Huang W, Li L, Myers JR, et al. ART: a next-generation sequencing read simulator[J]. Bioinformatics. 2012;28(4):593–4.View ArticlePubMedGoogle Scholar
- Chalitchagorn K, Shuangshoti S, Hourpai N, et al. Distinctive pattern of LINE-1 methylation level in normal tissues and the association with carcinogenesis[J]. Oncogene. 2004;23(54):8841–6.View ArticlePubMedGoogle Scholar
- Guo W, Fiziev P, Yan W, et al. BS-Seeker2: a versatile aligning pipeline for bisulfite sequencing data. BMC Genomics. 2013;14:774.View ArticlePubMedPubMed CentralGoogle Scholar
- Molaro A, Hodges E, Fang F, et al. Sperm methylation profiles reveal features of epigenetic inheritance and evolution in primates[J]. Cell. 2011;146(6):1029–41.View ArticlePubMedPubMed CentralGoogle Scholar
- Zou Q, Li XB, Jiang WR, et al. Survey of MapReduce frame operation in bioinformatics[J]. Brief Bioinform. 2014;15(4):637–47.View ArticlePubMedGoogle Scholar
- Li M, Yan X, Zhao Z, et al. VAliBS: a visual aligner for bisulfite sequences. ISBRA2016. A. Bourgeois et al. (Eds.): LNBI 9683, 307–308. DOI: 10.1007/978-3-319-38782-6.