Upload
others
View
5
Download
0
Embed Size (px)
Citation preview
Resource ReviewHBS-Tools for Hairpin Bisulfite Sequencing DataProcessing and Analysis
Ming-an Sun1 Karthik Raja Velmurugan12 David Keimig1 and Hehuang Xie123
1Epigenomics and Computational Biology Lab Virginia Bioinformatics Institute Virginia Tech Blacksburg VA 24061 USA2Interdisciplinary PhD Program in Genomics Bioinformatics and Computational Biology Virginia Tech Blacksburg VA 24061 USA3Department of Biological Sciences Virginia Tech Blacksburg VA 24061 USA
Correspondence should be addressed to Hehuang Xie davidxievtedu
Received 5 September 2015 Accepted 3 December 2015
Academic Editor David Corne
Copyright copy 2015 Ming-an Sun et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
The emerging genome-wide hairpin bisulfite sequencing (hairpin-BS-Seq) technique enables the determination of the methylationpattern for DNA double strands simultaneously Compared with traditional bisulfite sequencing (BS-Seq) techniques hairpin-BS-Seq can determine methylation fidelity and increase mapping efficiency However no computational tool has been designed for theanalysis of hairpin-BS-Seq data yet Here we present HBS-tools a set of command line based tools for the preprocessing mappingmethylation calling and summarizing of genome-wide hairpin-BS-Seq data It accepts paired-end hairpin-BS-Seq reads to recoverthe original (pre-bisulfite-converted) sequences using global alignment and then calls themethylation statuses for cytosines on bothDNA strands after mapping the original sequences to the reference genome After applying to hairpin-BS-Seq datasets we foundthat HBS-tools have a reduced mapping time and improved mapping efficiency compared with state-of-the-art mapping tools TheHBS-tools source scripts along with user guide and testing data are freely available for download
1 Introduction
During cell division DNAmethylation patterns are faithfullycopied from the parental to daughter strands by DNAmethyltransferase 1 [1 2] Although most cytosines at CpGdyads were found to be either symmetrically methylatedor completely unmethylated asymmetric DNA methylationat certain genomic loci has been found to be associatedwith stochastic methylation changes in normal tissues andcontribute to the epigenetic heterogeneity and eventually tophenotypic diversity [3] In addition increased asymmetricmethylationwas frequently observed in tumors with unstableepigenomes [4]
To determine the symmetry of CpG methylation Lairdand colleagues developed a hairpin bisulfite PCR techniqueto generatemethylation data for both complementary strandssimultaneously [5] Recently we implemented a genome-wide hairpin-BS-Seq technique to enable the assessmentof global methylation fidelity [6] In brief genomic DNAwas extracted and then sonicated into short fragments and
ligated to the biotinylated hairpin and Illumina sequencingadaptors simultaneously Following the streptavidin-captureand bisulfite PCR the fragments linked to both the hairpinadaptor and Illumina sequencing adaptor were amplifiedfor high-throughput paired-end sequencing Compared totraditional BS-Seq strategies hairpin-BS-Seq provides severaladvantages apart from assessing methylation inheritancefidelity (1) unlike traditional BS-Seq techniques which resultin reduced sequence complexity the possibility of recoveringthe original (pre-bisulfite-converted) sequence from hairpin-BS-Seq data to improve mapping efficiency (2) the ability toaccurately determine the SNPs including C-to-T conversion(3) the estimation of PCR andor sequencing errors byexamining the mismatches (excluding C-to-T and G-to-Amismatches which could result from bisulfite conversion)between read1 and read2
In the past years great efforts have been made to developexcellent algorithms and tools for the processing and analyz-ing of traditional BS-Seq data [7ndash10] but none for hairpin-BS-Seq data In this study we designed and implemented
Hindawi Publishing CorporationAdvances in BioinformaticsVolume 2015 Article ID 760423 4 pageshttpdxdoiorg1011552015760423
2 Advances in Bioinformatics
Table 1 Summary of the programs included in HBS-tools
Module name Function
hbs process Preprocessing of raw reads including bad quality bases trimming sequencing adaptor and hairpin adaptorremoval
hbs mapper Original sequence recovery mapping and SAM file outputhbs methylation extractor Extract and output methylation pattern from the SAM filehbs cg mlmf Summarize the methylation level and fidelity for covered CpG siteshbs ch ml Summarize the methylation level for covered non CpG sites
HBS-tools and compared them against other state-of-the-artmapping tools Our result indicated that HBS-tools have areduced mapping time and improved mapping efficiency
2 Software Description
HBS-tools include a set of scripts (implemented in PERL andC) for the processing and analysis of hairpin-BS-Seq data(Table 1) The functions of core modules are described asbelow
21 hbs process The hbs process is designed for the pro-cessing of hairpin-BS-Seq data It takes raw fastQ files fromhairpin-BS-Seq as input and integrates functions including(1) trimming bad quality residues from the input sequence(2) filtering hairpin adaptors (3) filtering sequencing adap-tors (4) discarding read pairs with any read shorter than thegiven threshold after (1)ndash(3) steps
22 hbs mapper The hbs mapper is the program for map-ping hairpin-BS-Seq reads to the reference genome andobtaining methylation calls subsequently Unlike previouslypublished tools which usually map bisulfite-converted readsto reference genomes after C-to-T and G-to-A conversion ofboth reads and reference genomes hbs mapper fully takesadvantage of the merits of hairpin-BS-Seq reads and utilizeda special recover-then-mapping strategy for read mappingIn brief it first recovers the original sequences after globalalignment of read1 and read2 with the Needleman-Wunschalgorithm using a modified scoring matrix which toleratesthe inconsistence between read1 and read2 probably dueto bisulfite conversion (eg C-to-T in read1 and G-to-A inread2) After trimming the overhangs for the alignment atthe two ends which may be due to different length of read1and read2 andor sequencing errors the recovered sequencesare mapped to the reference using Bowtie1 or Bowtie2 [11 12](Figure 1) Such amapping procedure overcomes the reducedsequence complexity which is evident for traditional BS-Seqand thus improves mapping efficiency
After global alignment of read1 and read2 the originalsequence is recovered by following four simple rules (a) a T inread1 and a C in read2 represent a C-to-T conversion duringbisulfite treatment and hence the original sequencemust havehad a C (b) a G in read1 and an A in read2 represent a G-to-A conversion and hence the original sequence must have
Original sequence recovery
Mapping original to reference
Read1 Z x x h Z Z x x
Z
Read1 Read2
Read2Methylation call
Bisulfite PCR
Genomic DNA
MethylatedUnmethylatedReference genome
Hairpin adaptorSequencing direction
Read1Read2
Original
Sequencing5998400
5998400
3998400
3998400
TCGTAGTAGTAACGTCGCAACAACAACGTCGCAGCAGCAACG
GATCGCAGCAGCAACGCA
GATCGCAGCAGCAACGCA GATCGCAGCAGCAACGCA
TCGTAGTAGTAACG TCGCAACAACAACG
TCGCAGCAGCAACG
TCGTAGTAGTAACG TCGCAACAACAACG
TCGCAGCAGCAACG
AGCGTCGTCGTTGC
TCGTAGTAGTAACG CGTTGTTGTTGCGAAGCATCATCATTGC GCAACAACAACGCT
Figure 1 Genome-wide hairpin bisulfite sequence generation andprocessing The flow chart begins with double stranded genomicDNA ligated to hairpin adapter Hairpin bisulfite PCR productsare sequenced from both ends HBS-tools accept these pair-endhairpin-BS-Seq reads to recover the original (pre-bisulfite con-verted) sequences The recovered sequences are aligned with thereference genome Methylation calls are obtained based on thesequence alignment between raw sequence reads and correspondinggenome sequences
had a G (c) when read1 and read2 have the same nucleotideit represents no modification and hence stays the same inthe recovered original sequence (d) when read1 and read2have different nucleotides that are not due to C-to-T or G-to-A conversion the one with the better quality score will bekept The recovered original sequence is then mapped to thereference genome using Bowtie1 or Bowtie2 [11 12] Havingtracked the reference genome fragment that corresponds tothe original sequence the raw read1 and read2 are comparedto the reference genome fragment to call the methylationstatuses for covered cytosines
The methylation calls and the alignment informationare generated in standard SAM format [13] The outputcontains information such as read ID chromosome genomic
Advances in Bioinformatics 3
position and methylation calls The methylation call stringis designed in a fashion so as to represent the methylationstatuses of cytosines in three possible contexts The smalland capital letters of ldquozrdquo ldquoxrdquo and ldquohrdquo are used to representthe unmethylated and methylated events at CpG CHG andCHH sites respectively The mapping output can be used forpostprocessing to extract methylation call information forindividual cytosines
23 hbs methylation extractor The hbs methylation ex-tractor takes the SAM file generated by hbs mapper asinput parses the methylation call strings and extracts themethylation statuses for the cytosines covered by hairpin-BS-Seq reads It provides the options to output methylationinformation for CpG and non-CpG contexts either separatelyor together
In the hbs mapper output each line represents themapping and methylation call for a sequence read In theextractor output each line contains the information for themethylation status of one cytosine covered by a sequenceread Apart from the read ID and methylation status of acytosine the extractor output also contains chromosomegenomic coordinate and strand information
24 hbs cg mlmf and hbs ch ml In addition to the methy-lation pattern obtained from each read we are also inter-ested in the methylation patterns for CpG dyads along thegenome Thus two simple yet useful scripts hbs cg mlmfand hbs ch ml were designed to summarize the methy-lation patterns for CpG and non-CpG sites respectivelyhbs cg mlmf takes the CpG methylation callings generatedfrom the hbs methylation extractor to calculate the methyla-tion level methylation fidelity and other related informationfor each CpG site determined Similarly hbs ch ml takes thenon CpG methylation calling result as input to calculate themethylation level for each covered non CpG siteThe outputsof two scripts can be used for further comparison betweendifferent samples
3 Software Performance
To test the performance ofHBS-tools we applied hbs mapperto hairpin-BS-Seq data [6] retrieved from NCBI SequenceRead Archive (SRA) database with accession numbersSRR919303 SRR919304 SRR919305 and SRR919306 Thesehairpin-BS-Seq data were generated for self-renewal mouseembryonic stem cells (ESCs) using the Illumina HiSeq2000 platform All reads are of 101 bp in length Thebad quality bases hairpin adaptor and sequencing adap-tor were trimmed using the hbs process and read pairsshorter than 50 bp after trimming were discarded Finallywe obtained 314M 318M 318M and 317M read pairsfor these four datasets respectively The reference genome(mm10) was downloaded from the UCSC genome browser[14]
The processed datasets were then used as the benchmarkfor the comparison between hbs mapper and Bismark [9]which is the most widely used aligner and methylation caller
for BS-Seq data Notably both hbs mapper and Bismarkused Bowtie as the engine for mapping The analysis wasconducted on a large-memory server of 12-core 290GHzIntel Xeon CPU that runs SUSE Linux operating systemFor unbiased comparison the same parameters (-119899 2 -119897 50)for Bowtie were used for hbs mapper and Bismark Read1and read2 were mapped separately using Bismark becausehairpin-BS-Seq data is different from traditional paired-endBS-Seq data and cannot be mapped using the paired-endmode by Bismark The result indicated that hbs mapper ismore time-efficient and could achieve improved mappingefficiency for hairpin-BS-Seq data
31 Running Time We compared the running times formap-ping hairpin-BS-Seq dataThe result showed that hbs mapperis more time-efficient compared with Bismark In averagehbs mapper uses around 358 less CPU hours comparedwith Bismark This is because hbs mapper maps the recov-ered original sequence to the reference genome using onethread while Bismark needs to map the C-to-T and G-to-A converted reads to the reference genome using two(directional DNA library) or four (nondirectional library)threads
32Mapping Efficiency Wefirst compared the percentages ofread pairsmapped by hbs mapper and Bismark For Bismarkread1 and read2 were first mapped separately to the referencegenome and then mapped read pairs were determinedas those with both read1 and read2 mapped to the samechromosome and are less than 50 bp away By doing it thisway around 444 of read pairs could bemapped by Bismark(Figure 2(a)) When using hbs mapper around 517 of readpairs could be mapped to reference (Figure 2(b)) We nextasked if more reads (read1 or read2) could be mapped bycombining hbs mapper and Bismark The result showed thatwhile around 558 (between 557 and 560) reads couldbemapped by using only Bismark 609 (between 608 and611) were mapped by combining the results of hbs mapperand Bismark
4 Discussion
Traditional BS-Seq cannot determine hemimethylation andalso suffers from mapping errors due to reduced DNAcomplexity In contrast hairpin-BS-Seq allows the determi-nation of methylation information from both DNA strandssimultaneously Therefore hairpin-BS-Seq could not only beused to assess methylation fidelity but also have the poten-tial to improve mapping efficiency by recovering originalsequences from the read pairs Here we described HBS-tools a set of programs specially designed for the analysisof genome-wide hairpin-BS-Seq data When applied to realhairpin-BS-Seq data the result indicated that HBS-tools aremore time-efficient and have improved mapping efficiencycompared with similar tools designed for traditional BS-Seqdata
4 Advances in Bioinformatics
0
5
10
15
20
SRR919303 SRR919304 SRR919305 SRR919306
CPU
tim
e (ho
urs)
Bismarkhbs_mapper
(a)
0
10
20
30
40
50
SRR919303 SRR919304 SRR919305 SRR919306
Map
ped
read
pai
rs (
)
Bismarkhbs_mapper
(b)
Figure 2 Comparison of the running times and mapping efficiencies between HBS-tools and Bismark (a) CPU time used by Bismark andhbs mapper for the mapping of public hairpin-BS-Seq datasets to reference (b) The percentage of hairpin-BS-Seq read pairs mapped byBismark and hbs mapper respectively
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work was supported by the VBI new faculty startup fundfor Hehuang Xie
References
[1] R Goyal R Reinhardt and A Jeltsch ldquoAccuracy of DNAmethylation pattern preservation by the Dnmt1 methyltrans-feraserdquoNucleic Acids Research vol 34 no 4 pp 1182ndash1188 2006
[2] G Vilkaitis I Suetake S Klimasauskas and S Tajima ldquoProces-sivemethylation of hemimethylated CpG sites bymouse Dnmt1DNA methyltransferaserdquo The Journal of Biological Chemistryvol 280 no 1 pp 64ndash72 2005
[3] H Xie M Wang A de Andrade et al ldquoGenome-wide quan-titative assessment of variation in DNA methylation patternsrdquoNucleic Acids Research vol 39 no 10 pp 4099ndash4108 2011
[4] C Shao M Lacey L Dubeau and M Ehrlich ldquoHemimethy-lation footprints of DNA demethylation in cancerrdquo Epigeneticsvol 4 no 3 pp 165ndash175 2009
[5] C D Laird N D Pleasant A D Clark et al ldquoHairpin-bisulfitePCR assessing epigenetic methylation patterns on complemen-tary strands of individual DNA moleculesrdquo Proceedings of theNational Academy of Sciences of the United States of Americavol 101 no 1 pp 204ndash209 2004
[6] L Zhao M-A Sun Z Li et al ldquoThe dynamics of DNAmethylation fidelity during mouse embryonic stem cell self-renewal and differentiationrdquo Genome Research vol 24 no 8pp 1296ndash1307 2014
[7] Y Xi and W Li ldquoBSMAP whole genome bisulfite sequenceMAPping programrdquo BMC Bioinformatics vol 10 article 2322009
[8] P-Y Chen S J Cokus and M Pellegrini ldquoBS Seeker precisemapping for bisulfite sequencingrdquo BMC Bioinformatics vol 11article 203 2010
[9] F Krueger and S R Andrews ldquoBismark a flexible aligner andmethylation caller for Bisulfite-Seq applicationsrdquo Bioinformat-ics vol 27 no 11 pp 1571ndash1572 2011
[10] J-Q Lim C Tennakoon G Li et al ldquoBatMeth improvedmapper for bisulfite sequencing reads on DNA methylationrdquoGenome Biology vol 13 no 10 article R82 2012
[11] B Langmead C Trapnell M Pop and S L Salzberg ldquoUltrafastandmemory-efficient alignment of short DNA sequences to thehuman genomerdquo Genome Biology vol 10 article R25 2009
[12] B Langmead and S L Salzberg ldquoFast gapped-read alignmentwith Bowtie 2rdquo Nature Methods vol 9 no 4 pp 357ndash359 2012
[13] H Li B Handsaker A Wysoker et al ldquoThe sequence align-mentmap format and SAMtoolsrdquoBioinformatics vol 25 no 16pp 2078ndash2079 2009
[14] P A Fujita B Rhead A S Zweig et al ldquoThe UCSC genomebrowser database update 2011rdquo Nucleic Acids Research vol 39no 1 pp D876ndashD882 2011
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Anatomy Research International
PeptidesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
International Journal of
Volume 2014
Zoology
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Molecular Biology International
GenomicsInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
BioinformaticsAdvances in
Marine BiologyJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Signal TransductionJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
BioMed Research International
Evolutionary BiologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Biochemistry Research International
ArchaeaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Genetics Research International
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Virolog y
Hindawi Publishing Corporationhttpwwwhindawicom
Nucleic AcidsJournal of
Volume 2014
Stem CellsInternational
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Enzyme Research
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Microbiology
2 Advances in Bioinformatics
Table 1 Summary of the programs included in HBS-tools
Module name Function
hbs process Preprocessing of raw reads including bad quality bases trimming sequencing adaptor and hairpin adaptorremoval
hbs mapper Original sequence recovery mapping and SAM file outputhbs methylation extractor Extract and output methylation pattern from the SAM filehbs cg mlmf Summarize the methylation level and fidelity for covered CpG siteshbs ch ml Summarize the methylation level for covered non CpG sites
HBS-tools and compared them against other state-of-the-artmapping tools Our result indicated that HBS-tools have areduced mapping time and improved mapping efficiency
2 Software Description
HBS-tools include a set of scripts (implemented in PERL andC) for the processing and analysis of hairpin-BS-Seq data(Table 1) The functions of core modules are described asbelow
21 hbs process The hbs process is designed for the pro-cessing of hairpin-BS-Seq data It takes raw fastQ files fromhairpin-BS-Seq as input and integrates functions including(1) trimming bad quality residues from the input sequence(2) filtering hairpin adaptors (3) filtering sequencing adap-tors (4) discarding read pairs with any read shorter than thegiven threshold after (1)ndash(3) steps
22 hbs mapper The hbs mapper is the program for map-ping hairpin-BS-Seq reads to the reference genome andobtaining methylation calls subsequently Unlike previouslypublished tools which usually map bisulfite-converted readsto reference genomes after C-to-T and G-to-A conversion ofboth reads and reference genomes hbs mapper fully takesadvantage of the merits of hairpin-BS-Seq reads and utilizeda special recover-then-mapping strategy for read mappingIn brief it first recovers the original sequences after globalalignment of read1 and read2 with the Needleman-Wunschalgorithm using a modified scoring matrix which toleratesthe inconsistence between read1 and read2 probably dueto bisulfite conversion (eg C-to-T in read1 and G-to-A inread2) After trimming the overhangs for the alignment atthe two ends which may be due to different length of read1and read2 andor sequencing errors the recovered sequencesare mapped to the reference using Bowtie1 or Bowtie2 [11 12](Figure 1) Such amapping procedure overcomes the reducedsequence complexity which is evident for traditional BS-Seqand thus improves mapping efficiency
After global alignment of read1 and read2 the originalsequence is recovered by following four simple rules (a) a T inread1 and a C in read2 represent a C-to-T conversion duringbisulfite treatment and hence the original sequencemust havehad a C (b) a G in read1 and an A in read2 represent a G-to-A conversion and hence the original sequence must have
Original sequence recovery
Mapping original to reference
Read1 Z x x h Z Z x x
Z
Read1 Read2
Read2Methylation call
Bisulfite PCR
Genomic DNA
MethylatedUnmethylatedReference genome
Hairpin adaptorSequencing direction
Read1Read2
Original
Sequencing5998400
5998400
3998400
3998400
TCGTAGTAGTAACGTCGCAACAACAACGTCGCAGCAGCAACG
GATCGCAGCAGCAACGCA
GATCGCAGCAGCAACGCA GATCGCAGCAGCAACGCA
TCGTAGTAGTAACG TCGCAACAACAACG
TCGCAGCAGCAACG
TCGTAGTAGTAACG TCGCAACAACAACG
TCGCAGCAGCAACG
AGCGTCGTCGTTGC
TCGTAGTAGTAACG CGTTGTTGTTGCGAAGCATCATCATTGC GCAACAACAACGCT
Figure 1 Genome-wide hairpin bisulfite sequence generation andprocessing The flow chart begins with double stranded genomicDNA ligated to hairpin adapter Hairpin bisulfite PCR productsare sequenced from both ends HBS-tools accept these pair-endhairpin-BS-Seq reads to recover the original (pre-bisulfite con-verted) sequences The recovered sequences are aligned with thereference genome Methylation calls are obtained based on thesequence alignment between raw sequence reads and correspondinggenome sequences
had a G (c) when read1 and read2 have the same nucleotideit represents no modification and hence stays the same inthe recovered original sequence (d) when read1 and read2have different nucleotides that are not due to C-to-T or G-to-A conversion the one with the better quality score will bekept The recovered original sequence is then mapped to thereference genome using Bowtie1 or Bowtie2 [11 12] Havingtracked the reference genome fragment that corresponds tothe original sequence the raw read1 and read2 are comparedto the reference genome fragment to call the methylationstatuses for covered cytosines
The methylation calls and the alignment informationare generated in standard SAM format [13] The outputcontains information such as read ID chromosome genomic
Advances in Bioinformatics 3
position and methylation calls The methylation call stringis designed in a fashion so as to represent the methylationstatuses of cytosines in three possible contexts The smalland capital letters of ldquozrdquo ldquoxrdquo and ldquohrdquo are used to representthe unmethylated and methylated events at CpG CHG andCHH sites respectively The mapping output can be used forpostprocessing to extract methylation call information forindividual cytosines
23 hbs methylation extractor The hbs methylation ex-tractor takes the SAM file generated by hbs mapper asinput parses the methylation call strings and extracts themethylation statuses for the cytosines covered by hairpin-BS-Seq reads It provides the options to output methylationinformation for CpG and non-CpG contexts either separatelyor together
In the hbs mapper output each line represents themapping and methylation call for a sequence read In theextractor output each line contains the information for themethylation status of one cytosine covered by a sequenceread Apart from the read ID and methylation status of acytosine the extractor output also contains chromosomegenomic coordinate and strand information
24 hbs cg mlmf and hbs ch ml In addition to the methy-lation pattern obtained from each read we are also inter-ested in the methylation patterns for CpG dyads along thegenome Thus two simple yet useful scripts hbs cg mlmfand hbs ch ml were designed to summarize the methy-lation patterns for CpG and non-CpG sites respectivelyhbs cg mlmf takes the CpG methylation callings generatedfrom the hbs methylation extractor to calculate the methyla-tion level methylation fidelity and other related informationfor each CpG site determined Similarly hbs ch ml takes thenon CpG methylation calling result as input to calculate themethylation level for each covered non CpG siteThe outputsof two scripts can be used for further comparison betweendifferent samples
3 Software Performance
To test the performance ofHBS-tools we applied hbs mapperto hairpin-BS-Seq data [6] retrieved from NCBI SequenceRead Archive (SRA) database with accession numbersSRR919303 SRR919304 SRR919305 and SRR919306 Thesehairpin-BS-Seq data were generated for self-renewal mouseembryonic stem cells (ESCs) using the Illumina HiSeq2000 platform All reads are of 101 bp in length Thebad quality bases hairpin adaptor and sequencing adap-tor were trimmed using the hbs process and read pairsshorter than 50 bp after trimming were discarded Finallywe obtained 314M 318M 318M and 317M read pairsfor these four datasets respectively The reference genome(mm10) was downloaded from the UCSC genome browser[14]
The processed datasets were then used as the benchmarkfor the comparison between hbs mapper and Bismark [9]which is the most widely used aligner and methylation caller
for BS-Seq data Notably both hbs mapper and Bismarkused Bowtie as the engine for mapping The analysis wasconducted on a large-memory server of 12-core 290GHzIntel Xeon CPU that runs SUSE Linux operating systemFor unbiased comparison the same parameters (-119899 2 -119897 50)for Bowtie were used for hbs mapper and Bismark Read1and read2 were mapped separately using Bismark becausehairpin-BS-Seq data is different from traditional paired-endBS-Seq data and cannot be mapped using the paired-endmode by Bismark The result indicated that hbs mapper ismore time-efficient and could achieve improved mappingefficiency for hairpin-BS-Seq data
31 Running Time We compared the running times formap-ping hairpin-BS-Seq dataThe result showed that hbs mapperis more time-efficient compared with Bismark In averagehbs mapper uses around 358 less CPU hours comparedwith Bismark This is because hbs mapper maps the recov-ered original sequence to the reference genome using onethread while Bismark needs to map the C-to-T and G-to-A converted reads to the reference genome using two(directional DNA library) or four (nondirectional library)threads
32Mapping Efficiency Wefirst compared the percentages ofread pairsmapped by hbs mapper and Bismark For Bismarkread1 and read2 were first mapped separately to the referencegenome and then mapped read pairs were determinedas those with both read1 and read2 mapped to the samechromosome and are less than 50 bp away By doing it thisway around 444 of read pairs could bemapped by Bismark(Figure 2(a)) When using hbs mapper around 517 of readpairs could be mapped to reference (Figure 2(b)) We nextasked if more reads (read1 or read2) could be mapped bycombining hbs mapper and Bismark The result showed thatwhile around 558 (between 557 and 560) reads couldbemapped by using only Bismark 609 (between 608 and611) were mapped by combining the results of hbs mapperand Bismark
4 Discussion
Traditional BS-Seq cannot determine hemimethylation andalso suffers from mapping errors due to reduced DNAcomplexity In contrast hairpin-BS-Seq allows the determi-nation of methylation information from both DNA strandssimultaneously Therefore hairpin-BS-Seq could not only beused to assess methylation fidelity but also have the poten-tial to improve mapping efficiency by recovering originalsequences from the read pairs Here we described HBS-tools a set of programs specially designed for the analysisof genome-wide hairpin-BS-Seq data When applied to realhairpin-BS-Seq data the result indicated that HBS-tools aremore time-efficient and have improved mapping efficiencycompared with similar tools designed for traditional BS-Seqdata
4 Advances in Bioinformatics
0
5
10
15
20
SRR919303 SRR919304 SRR919305 SRR919306
CPU
tim
e (ho
urs)
Bismarkhbs_mapper
(a)
0
10
20
30
40
50
SRR919303 SRR919304 SRR919305 SRR919306
Map
ped
read
pai
rs (
)
Bismarkhbs_mapper
(b)
Figure 2 Comparison of the running times and mapping efficiencies between HBS-tools and Bismark (a) CPU time used by Bismark andhbs mapper for the mapping of public hairpin-BS-Seq datasets to reference (b) The percentage of hairpin-BS-Seq read pairs mapped byBismark and hbs mapper respectively
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work was supported by the VBI new faculty startup fundfor Hehuang Xie
References
[1] R Goyal R Reinhardt and A Jeltsch ldquoAccuracy of DNAmethylation pattern preservation by the Dnmt1 methyltrans-feraserdquoNucleic Acids Research vol 34 no 4 pp 1182ndash1188 2006
[2] G Vilkaitis I Suetake S Klimasauskas and S Tajima ldquoProces-sivemethylation of hemimethylated CpG sites bymouse Dnmt1DNA methyltransferaserdquo The Journal of Biological Chemistryvol 280 no 1 pp 64ndash72 2005
[3] H Xie M Wang A de Andrade et al ldquoGenome-wide quan-titative assessment of variation in DNA methylation patternsrdquoNucleic Acids Research vol 39 no 10 pp 4099ndash4108 2011
[4] C Shao M Lacey L Dubeau and M Ehrlich ldquoHemimethy-lation footprints of DNA demethylation in cancerrdquo Epigeneticsvol 4 no 3 pp 165ndash175 2009
[5] C D Laird N D Pleasant A D Clark et al ldquoHairpin-bisulfitePCR assessing epigenetic methylation patterns on complemen-tary strands of individual DNA moleculesrdquo Proceedings of theNational Academy of Sciences of the United States of Americavol 101 no 1 pp 204ndash209 2004
[6] L Zhao M-A Sun Z Li et al ldquoThe dynamics of DNAmethylation fidelity during mouse embryonic stem cell self-renewal and differentiationrdquo Genome Research vol 24 no 8pp 1296ndash1307 2014
[7] Y Xi and W Li ldquoBSMAP whole genome bisulfite sequenceMAPping programrdquo BMC Bioinformatics vol 10 article 2322009
[8] P-Y Chen S J Cokus and M Pellegrini ldquoBS Seeker precisemapping for bisulfite sequencingrdquo BMC Bioinformatics vol 11article 203 2010
[9] F Krueger and S R Andrews ldquoBismark a flexible aligner andmethylation caller for Bisulfite-Seq applicationsrdquo Bioinformat-ics vol 27 no 11 pp 1571ndash1572 2011
[10] J-Q Lim C Tennakoon G Li et al ldquoBatMeth improvedmapper for bisulfite sequencing reads on DNA methylationrdquoGenome Biology vol 13 no 10 article R82 2012
[11] B Langmead C Trapnell M Pop and S L Salzberg ldquoUltrafastandmemory-efficient alignment of short DNA sequences to thehuman genomerdquo Genome Biology vol 10 article R25 2009
[12] B Langmead and S L Salzberg ldquoFast gapped-read alignmentwith Bowtie 2rdquo Nature Methods vol 9 no 4 pp 357ndash359 2012
[13] H Li B Handsaker A Wysoker et al ldquoThe sequence align-mentmap format and SAMtoolsrdquoBioinformatics vol 25 no 16pp 2078ndash2079 2009
[14] P A Fujita B Rhead A S Zweig et al ldquoThe UCSC genomebrowser database update 2011rdquo Nucleic Acids Research vol 39no 1 pp D876ndashD882 2011
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Anatomy Research International
PeptidesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
International Journal of
Volume 2014
Zoology
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Molecular Biology International
GenomicsInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
BioinformaticsAdvances in
Marine BiologyJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Signal TransductionJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
BioMed Research International
Evolutionary BiologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Biochemistry Research International
ArchaeaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Genetics Research International
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Virolog y
Hindawi Publishing Corporationhttpwwwhindawicom
Nucleic AcidsJournal of
Volume 2014
Stem CellsInternational
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Enzyme Research
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Microbiology
Advances in Bioinformatics 3
position and methylation calls The methylation call stringis designed in a fashion so as to represent the methylationstatuses of cytosines in three possible contexts The smalland capital letters of ldquozrdquo ldquoxrdquo and ldquohrdquo are used to representthe unmethylated and methylated events at CpG CHG andCHH sites respectively The mapping output can be used forpostprocessing to extract methylation call information forindividual cytosines
23 hbs methylation extractor The hbs methylation ex-tractor takes the SAM file generated by hbs mapper asinput parses the methylation call strings and extracts themethylation statuses for the cytosines covered by hairpin-BS-Seq reads It provides the options to output methylationinformation for CpG and non-CpG contexts either separatelyor together
In the hbs mapper output each line represents themapping and methylation call for a sequence read In theextractor output each line contains the information for themethylation status of one cytosine covered by a sequenceread Apart from the read ID and methylation status of acytosine the extractor output also contains chromosomegenomic coordinate and strand information
24 hbs cg mlmf and hbs ch ml In addition to the methy-lation pattern obtained from each read we are also inter-ested in the methylation patterns for CpG dyads along thegenome Thus two simple yet useful scripts hbs cg mlmfand hbs ch ml were designed to summarize the methy-lation patterns for CpG and non-CpG sites respectivelyhbs cg mlmf takes the CpG methylation callings generatedfrom the hbs methylation extractor to calculate the methyla-tion level methylation fidelity and other related informationfor each CpG site determined Similarly hbs ch ml takes thenon CpG methylation calling result as input to calculate themethylation level for each covered non CpG siteThe outputsof two scripts can be used for further comparison betweendifferent samples
3 Software Performance
To test the performance ofHBS-tools we applied hbs mapperto hairpin-BS-Seq data [6] retrieved from NCBI SequenceRead Archive (SRA) database with accession numbersSRR919303 SRR919304 SRR919305 and SRR919306 Thesehairpin-BS-Seq data were generated for self-renewal mouseembryonic stem cells (ESCs) using the Illumina HiSeq2000 platform All reads are of 101 bp in length Thebad quality bases hairpin adaptor and sequencing adap-tor were trimmed using the hbs process and read pairsshorter than 50 bp after trimming were discarded Finallywe obtained 314M 318M 318M and 317M read pairsfor these four datasets respectively The reference genome(mm10) was downloaded from the UCSC genome browser[14]
The processed datasets were then used as the benchmarkfor the comparison between hbs mapper and Bismark [9]which is the most widely used aligner and methylation caller
for BS-Seq data Notably both hbs mapper and Bismarkused Bowtie as the engine for mapping The analysis wasconducted on a large-memory server of 12-core 290GHzIntel Xeon CPU that runs SUSE Linux operating systemFor unbiased comparison the same parameters (-119899 2 -119897 50)for Bowtie were used for hbs mapper and Bismark Read1and read2 were mapped separately using Bismark becausehairpin-BS-Seq data is different from traditional paired-endBS-Seq data and cannot be mapped using the paired-endmode by Bismark The result indicated that hbs mapper ismore time-efficient and could achieve improved mappingefficiency for hairpin-BS-Seq data
31 Running Time We compared the running times formap-ping hairpin-BS-Seq dataThe result showed that hbs mapperis more time-efficient compared with Bismark In averagehbs mapper uses around 358 less CPU hours comparedwith Bismark This is because hbs mapper maps the recov-ered original sequence to the reference genome using onethread while Bismark needs to map the C-to-T and G-to-A converted reads to the reference genome using two(directional DNA library) or four (nondirectional library)threads
32Mapping Efficiency Wefirst compared the percentages ofread pairsmapped by hbs mapper and Bismark For Bismarkread1 and read2 were first mapped separately to the referencegenome and then mapped read pairs were determinedas those with both read1 and read2 mapped to the samechromosome and are less than 50 bp away By doing it thisway around 444 of read pairs could bemapped by Bismark(Figure 2(a)) When using hbs mapper around 517 of readpairs could be mapped to reference (Figure 2(b)) We nextasked if more reads (read1 or read2) could be mapped bycombining hbs mapper and Bismark The result showed thatwhile around 558 (between 557 and 560) reads couldbemapped by using only Bismark 609 (between 608 and611) were mapped by combining the results of hbs mapperand Bismark
4 Discussion
Traditional BS-Seq cannot determine hemimethylation andalso suffers from mapping errors due to reduced DNAcomplexity In contrast hairpin-BS-Seq allows the determi-nation of methylation information from both DNA strandssimultaneously Therefore hairpin-BS-Seq could not only beused to assess methylation fidelity but also have the poten-tial to improve mapping efficiency by recovering originalsequences from the read pairs Here we described HBS-tools a set of programs specially designed for the analysisof genome-wide hairpin-BS-Seq data When applied to realhairpin-BS-Seq data the result indicated that HBS-tools aremore time-efficient and have improved mapping efficiencycompared with similar tools designed for traditional BS-Seqdata
4 Advances in Bioinformatics
0
5
10
15
20
SRR919303 SRR919304 SRR919305 SRR919306
CPU
tim
e (ho
urs)
Bismarkhbs_mapper
(a)
0
10
20
30
40
50
SRR919303 SRR919304 SRR919305 SRR919306
Map
ped
read
pai
rs (
)
Bismarkhbs_mapper
(b)
Figure 2 Comparison of the running times and mapping efficiencies between HBS-tools and Bismark (a) CPU time used by Bismark andhbs mapper for the mapping of public hairpin-BS-Seq datasets to reference (b) The percentage of hairpin-BS-Seq read pairs mapped byBismark and hbs mapper respectively
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work was supported by the VBI new faculty startup fundfor Hehuang Xie
References
[1] R Goyal R Reinhardt and A Jeltsch ldquoAccuracy of DNAmethylation pattern preservation by the Dnmt1 methyltrans-feraserdquoNucleic Acids Research vol 34 no 4 pp 1182ndash1188 2006
[2] G Vilkaitis I Suetake S Klimasauskas and S Tajima ldquoProces-sivemethylation of hemimethylated CpG sites bymouse Dnmt1DNA methyltransferaserdquo The Journal of Biological Chemistryvol 280 no 1 pp 64ndash72 2005
[3] H Xie M Wang A de Andrade et al ldquoGenome-wide quan-titative assessment of variation in DNA methylation patternsrdquoNucleic Acids Research vol 39 no 10 pp 4099ndash4108 2011
[4] C Shao M Lacey L Dubeau and M Ehrlich ldquoHemimethy-lation footprints of DNA demethylation in cancerrdquo Epigeneticsvol 4 no 3 pp 165ndash175 2009
[5] C D Laird N D Pleasant A D Clark et al ldquoHairpin-bisulfitePCR assessing epigenetic methylation patterns on complemen-tary strands of individual DNA moleculesrdquo Proceedings of theNational Academy of Sciences of the United States of Americavol 101 no 1 pp 204ndash209 2004
[6] L Zhao M-A Sun Z Li et al ldquoThe dynamics of DNAmethylation fidelity during mouse embryonic stem cell self-renewal and differentiationrdquo Genome Research vol 24 no 8pp 1296ndash1307 2014
[7] Y Xi and W Li ldquoBSMAP whole genome bisulfite sequenceMAPping programrdquo BMC Bioinformatics vol 10 article 2322009
[8] P-Y Chen S J Cokus and M Pellegrini ldquoBS Seeker precisemapping for bisulfite sequencingrdquo BMC Bioinformatics vol 11article 203 2010
[9] F Krueger and S R Andrews ldquoBismark a flexible aligner andmethylation caller for Bisulfite-Seq applicationsrdquo Bioinformat-ics vol 27 no 11 pp 1571ndash1572 2011
[10] J-Q Lim C Tennakoon G Li et al ldquoBatMeth improvedmapper for bisulfite sequencing reads on DNA methylationrdquoGenome Biology vol 13 no 10 article R82 2012
[11] B Langmead C Trapnell M Pop and S L Salzberg ldquoUltrafastandmemory-efficient alignment of short DNA sequences to thehuman genomerdquo Genome Biology vol 10 article R25 2009
[12] B Langmead and S L Salzberg ldquoFast gapped-read alignmentwith Bowtie 2rdquo Nature Methods vol 9 no 4 pp 357ndash359 2012
[13] H Li B Handsaker A Wysoker et al ldquoThe sequence align-mentmap format and SAMtoolsrdquoBioinformatics vol 25 no 16pp 2078ndash2079 2009
[14] P A Fujita B Rhead A S Zweig et al ldquoThe UCSC genomebrowser database update 2011rdquo Nucleic Acids Research vol 39no 1 pp D876ndashD882 2011
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Anatomy Research International
PeptidesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
International Journal of
Volume 2014
Zoology
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Molecular Biology International
GenomicsInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
BioinformaticsAdvances in
Marine BiologyJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Signal TransductionJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
BioMed Research International
Evolutionary BiologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Biochemistry Research International
ArchaeaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Genetics Research International
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Virolog y
Hindawi Publishing Corporationhttpwwwhindawicom
Nucleic AcidsJournal of
Volume 2014
Stem CellsInternational
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Enzyme Research
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Microbiology
4 Advances in Bioinformatics
0
5
10
15
20
SRR919303 SRR919304 SRR919305 SRR919306
CPU
tim
e (ho
urs)
Bismarkhbs_mapper
(a)
0
10
20
30
40
50
SRR919303 SRR919304 SRR919305 SRR919306
Map
ped
read
pai
rs (
)
Bismarkhbs_mapper
(b)
Figure 2 Comparison of the running times and mapping efficiencies between HBS-tools and Bismark (a) CPU time used by Bismark andhbs mapper for the mapping of public hairpin-BS-Seq datasets to reference (b) The percentage of hairpin-BS-Seq read pairs mapped byBismark and hbs mapper respectively
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work was supported by the VBI new faculty startup fundfor Hehuang Xie
References
[1] R Goyal R Reinhardt and A Jeltsch ldquoAccuracy of DNAmethylation pattern preservation by the Dnmt1 methyltrans-feraserdquoNucleic Acids Research vol 34 no 4 pp 1182ndash1188 2006
[2] G Vilkaitis I Suetake S Klimasauskas and S Tajima ldquoProces-sivemethylation of hemimethylated CpG sites bymouse Dnmt1DNA methyltransferaserdquo The Journal of Biological Chemistryvol 280 no 1 pp 64ndash72 2005
[3] H Xie M Wang A de Andrade et al ldquoGenome-wide quan-titative assessment of variation in DNA methylation patternsrdquoNucleic Acids Research vol 39 no 10 pp 4099ndash4108 2011
[4] C Shao M Lacey L Dubeau and M Ehrlich ldquoHemimethy-lation footprints of DNA demethylation in cancerrdquo Epigeneticsvol 4 no 3 pp 165ndash175 2009
[5] C D Laird N D Pleasant A D Clark et al ldquoHairpin-bisulfitePCR assessing epigenetic methylation patterns on complemen-tary strands of individual DNA moleculesrdquo Proceedings of theNational Academy of Sciences of the United States of Americavol 101 no 1 pp 204ndash209 2004
[6] L Zhao M-A Sun Z Li et al ldquoThe dynamics of DNAmethylation fidelity during mouse embryonic stem cell self-renewal and differentiationrdquo Genome Research vol 24 no 8pp 1296ndash1307 2014
[7] Y Xi and W Li ldquoBSMAP whole genome bisulfite sequenceMAPping programrdquo BMC Bioinformatics vol 10 article 2322009
[8] P-Y Chen S J Cokus and M Pellegrini ldquoBS Seeker precisemapping for bisulfite sequencingrdquo BMC Bioinformatics vol 11article 203 2010
[9] F Krueger and S R Andrews ldquoBismark a flexible aligner andmethylation caller for Bisulfite-Seq applicationsrdquo Bioinformat-ics vol 27 no 11 pp 1571ndash1572 2011
[10] J-Q Lim C Tennakoon G Li et al ldquoBatMeth improvedmapper for bisulfite sequencing reads on DNA methylationrdquoGenome Biology vol 13 no 10 article R82 2012
[11] B Langmead C Trapnell M Pop and S L Salzberg ldquoUltrafastandmemory-efficient alignment of short DNA sequences to thehuman genomerdquo Genome Biology vol 10 article R25 2009
[12] B Langmead and S L Salzberg ldquoFast gapped-read alignmentwith Bowtie 2rdquo Nature Methods vol 9 no 4 pp 357ndash359 2012
[13] H Li B Handsaker A Wysoker et al ldquoThe sequence align-mentmap format and SAMtoolsrdquoBioinformatics vol 25 no 16pp 2078ndash2079 2009
[14] P A Fujita B Rhead A S Zweig et al ldquoThe UCSC genomebrowser database update 2011rdquo Nucleic Acids Research vol 39no 1 pp D876ndashD882 2011
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Anatomy Research International
PeptidesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
International Journal of
Volume 2014
Zoology
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Molecular Biology International
GenomicsInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
BioinformaticsAdvances in
Marine BiologyJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Signal TransductionJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
BioMed Research International
Evolutionary BiologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Biochemistry Research International
ArchaeaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Genetics Research International
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Virolog y
Hindawi Publishing Corporationhttpwwwhindawicom
Nucleic AcidsJournal of
Volume 2014
Stem CellsInternational
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Enzyme Research
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Microbiology
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Anatomy Research International
PeptidesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
International Journal of
Volume 2014
Zoology
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Molecular Biology International
GenomicsInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
BioinformaticsAdvances in
Marine BiologyJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Signal TransductionJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
BioMed Research International
Evolutionary BiologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Biochemistry Research International
ArchaeaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Genetics Research International
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Virolog y
Hindawi Publishing Corporationhttpwwwhindawicom
Nucleic AcidsJournal of
Volume 2014
Stem CellsInternational
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Enzyme Research
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Microbiology