Development of genome wide insertion deletion markers in rice based on graphic pipeline platform

Development of genome-wide insertion/deletion markers in rice based on graphic pipeline platformYang L €u1†,Xiao Cui 1†,Rui Li 1,Piaopiao Huang 1,Jie Zong 2,Danqing Yao 3,Gang Li 3,Dabing Zhang 1and Zheng Yuan 1*1State Key Laboratory of Hybrid Rice,School of Life Sciences and Biotechnology,Shanghai Jiao Tong University,Shanghai 200240,China,2NovelBioinformatics Company,Shanghai 200240,China,3Shanghai Agricultural Technology Extension and Service Center,Shanghai 201103,China.†These authors contribute equally to this work.*Correspondence:zyuan@Abstract DNA markers play important roles in plant breed-ing and genetics.The Insertion/Deletion (InDel)marker is one kind of co-dominant DNA markers widely used due to its low cost and high precision.However,the canonical way of searching for InDel markers is time-consuming and labor-intensive.We developed an end-to-end computational solution (InDel Markers Development Platform,IMDP)to identify genome-wide InDel markers under a graphic pipeline environment.IMDP constitutes assembled genome sequen-ces alignment pipeline (AGA-pipe)and next-generation re-sequencing data mapping pipeline (NGS-pipe).With AGA-pipe we are able to identify 12,944markers between the genome of rice cultivars Nipponbare and ing NGS-pipe,we reported 34,794InDels from re-sequencing data of rice cultivars Wu-Yun-Geng7and bining AGA-pipe and NGS-pipe,we developed 205,659InDels in eight japonica and nine indica cultivars and 2,681InDels showed a subgroup-speci fic pattern.Polymerase chain reaction (PCR)analysis of subgroup-speci fic markers indicated that the precision reached 90%(86of 95).Finally,to make them available to the public,we have integrated the InDels/markers information into a website (Rice InDel Marker Database,RIMD,http://202.120.45.71/).The application of IMDP in rice will facilitate ef ficiency for development of genome-wide InDel markers,in addition it can be used in other species with reference genome sequences and NGS data.Keywords:Genetic polymorphism;genome alignment;InDel marker;molecular breeding;next-generation sequencingCitation:L €uY,Cui X,Li R,Huang P,Zong J,Yao D,Li G,Zhang D,Yuan Z (2015)Development of genome-wide insertion/deletion markers in rice based on graphic pipeline platform.J Integr Plant Biol XX:XX –XX doi:10.1111/jipb.12354Edited by:Li-Jia Qu,Peking University,China Received Dec.18,2014;Accepted Mar.19,2015Available online on Mar.21,2015at /journal/jipb©2015Institute of Botany,Chinese Academy of SciencesINTRODUCTIONRice (Oryza sativa L.)is the most widely consumed food staple for humans (Goff et al.2002).The rapid growth of world population coupled with a decrease in clean water and arable land resources,have urged plant scientists to place most of their efforts on the study of increasing yield of rice and other crops,during which molecular genetic techniques such as molecular marker-assisted selection (MAS)have been widely used (Mohan et al.1997).Successful MAS dependent crop breeding relies greatly on functional genomic analysis and application of agronomically important genes (Collard and Mackill 2008),as well as ef ficient screening assays based on DNA polymorphism detection (Collard et al.2005).Molecular markers have been developed rapidly during the past three decades,from hybridization-based molecular markers,such as restriction fragment length polymorphism (RFLP)(Botstein et al.1980),to polymerase chain reaction (PCR)-based molecular markers such as simple sequence repeat (SSR)(Powell et al.1996),random ampli fied poly-morphic DNA (RAPD)(Williams et al.1990),insertion/deletion markers (InDels)(Rickert et al.2003;Shen et al.2004),and to sequencing-based molecular markers such as single-nucleo-tide polymorphisms (SNPs)(Rafalski 2002).SNPs could also be transformed to cleave ampli fied polymorphism sequence (CAPS)(Konieczny and Ausubel 1993)and derived cleaved ampli fied polymorphic sequence (dCAPS)(Neff et al.1998),both of which were based on PCR and subsequent restriction enzyme digestion.All of these markers have been proved useful tools for species identi fication,genetic variation studies and population structure analysis,phylogenetic analysis,genetic mapping,and MAS (McCouch et al.1997).Given the drawbacks of SNPs and its derived markers such as high-cost or special equipment requirement,PCR-based markers (RAPD,SSR and InDels)are more widely used than SNPs in research and breeding.SSR markers are more widely used than RAPD because of convenient detection,moderate cost,high reproducibility,and multi-allele separation capa-bility (Powell et al.1996),but are limited due to insuf ficient genome abundance and coverage as well as ambiguities in complex and heterogeneous mutation patterns (PAN et al.2008;Liu et al.2013).The InDel marker,one kind of co-dominant PCR-based markers,has a higher order of magnitude than SSRs and displays no ambiguity under the same detection condition.Molecular markers are useful tools in research and breeding,but the high cost of marker development and marker application is the major limitation for its application (Collard and Mackill 2008).The genome-wide polymorphisms analysis and development of reliable molecular markers are time-consuming and labor-intensive,which also limit the application of MAS enormously.The application of next-Free AccessNew Technologygeneration sequencing(NGS)technologies have allowed the discovery of a large number of genome-wide DNA poly-morphisms,such as SNPs and InDels in relatively shorter time (Varshney et al.2009),which reduce the cost of marker development(Gao et al.2012).As the top two largest polymorphic mutations in the genome,the high density of SNP and InDel polymorphisms make them more suitable for genome-wide marker development(PAN et al.2008).In2009, the International Rice Functional Genomics Consortium initiated a rice sequence analysis program using20japonica/ indica cultivars and identified160,000SNPs(McNally et al. 2009).Huang et al.identified approximately3.6million SNPs and performed genome-wide association studies(GWAS)for 14agronomic traits via sequencing517rice landraces(Huang et al.2010).The application of SNPs genotyping analysis, however,is still not practical for most research laboratories due to its high cost and special equipment requirement(Liu et al.2013).In contrast,InDel markers are low-cost and technically easier to use for most laboratories.Therefore, genome-wide discovery of InDel polymorphisms offers a practical approach for basic research and MAS.With the continuous release of assembled and re-sequenced genomes, InDel markers have been developed in several plant species such as Arabidopsis(Hou et al.2010;Pacurar et al.2012), Brassica rapa(Liu et al.2013),and rice(Shen et al.2004;Arai-Kichise et al.2011).Despite many successful cases being reported,less time-and labor-consuming tools are in urgent need to be developed for discovery of genome-wide InDel markers.In this study,we developed an end-to-end computational solution(InDel Marker Development Platform,IMDP)to identify genome-wide InDel markers under Laboratory of Neuro Imaging(LONI)Pipeline Processing Environment(Rex et al.2003)for both assembled genome sequences and NGS ing assembled genome sequences pipeline (AGA-pipe)and next-generation re-sequencing data pipeline (NGS-pipe)of IMDP,Nipponbare/93-11and Wu-Yun-Geng7/ Guang-La-Ai4were analyzed for InDel markers development separately.Furthermore,we unified AGA-pipe and NGS-pipe and developed2,681subgroup InDels between eight japonica cultivars and nine indica rmation of all the InDels/ markers was then constructed into a user-friendly Rice InDel Markers Database(RIMD,http://202.120.45.71/),which would benefit the rice genomic research community in future studies and molecular breeding.The IMDP we developed in this study could also be applied in other species with assembled reference genome.RESULTSOverview of IMDPIMDP was designed with two process steps:credible InDels calling and automatic high quality primer design with e-PCR specificity validation.Depending on the nature of available genomic data(complete genome sequence or NGS data), two strategies for InDels calling were used in the InDel calling step:genome alignment(AGA-pipe)and NGS reads mapping(NGS-pipe).Both of AGA-pipe and NGS-pipe provide end-to-end solutions for InDel markers develop-ment,which differ in InDel calling processes but use the same modules in primer design and e-PCR validation (Figure1A).AGA-pipe used MUMmer3to obtain homologous regions, and sequence alignment of two assembled chromosomes and the output was subsequently processed by several scripts to identify optimal InDels for primer design and e-PCR validation. For rice cultivars without assembled genome sequences, NGS-pipe could be used to map re-sequencing data to reference genome and used it as a‘bridge’to identify InDel polymorphisms(Figure S1)(Liu et al.2013).NGS-pipe contains modules for trimming,sequence quality assessment,reads mapping,mapping quality assessment,InDel calling,primer design and e-PCR validation.In particular,NGS-pipe used Pindel for InDel detection,which excels in intermediate and long InDels detection that are suitable for PCR-based marker development.In the primer design module,Primer3program was used to automatically select primers based on user-specified criteria (Table S1),such as primer size,GC content,melting temper-ature,PCR product size,and others(Rozen and Skaletsky 2000).Followed by Primer3,MFEprimer2(e-PCR,computer simulations of theoretical PCR results)was used to evaluate the primers considering the conditions such as number of binding sites of the primers in the target sequence,stability at the30-end of the primer,melting temperature,GC content and several other criteria(Qu et al.2009).The chance of PCR failure or mis-priming due to off target matches in genome could be therefore greatly reduced.Finally,all IMDP models were wrapped into LONI processing pipeline environment (Figure1B)and could be easily installed(See Materials and Methods).The pipeline environment provides graphic inter-face,allowing bench scientists to set-up the analysis project, run the analysis and monitor the process with ease.InDel markers development using AGA-pipeUsing AGA-pipe of IMDP,we analyzed InDel polymorphisms in two inbred rice cultivars,japonica cv.Nipponbare and indica cv.93-11.The genomic sequence alignment indicated good collinearity between Nipponbare and93-11genome (Figure S2).In total,we identified90,548md-InDels in Nipponbare against93-11(Nipponbare/93-11)genome by get-indels.pl program.Among them,36,066hq-InDels were selected for PCR primer design using Primer3,followed by primer validation by MFEprimer2.Finally,12,944markers were obtained and could be used for further PCR and polyacryla-mide gel electrophoresis(PAGE)assay.The density of Nipponbare/93-11hq-InDels markers was3.7per BAC,33.4 per1Mb,which covered86.6%BACs of Nipponbare genome (2,998out of3,463;Table1).97.7%hq-InDel markers in Nipponbare/93-11were6–50bp in length(Figure S3;Table S2), which could be easily detected by general PCR and PAGE assay.In agreement with other InDel studies(Shen et al.2004; Arai-Kichise et al.2011),the distribution of InDel markers we found in Nipponbare/93-11was also unevenly distributed across the Nipponbare chromosomes,especially in or near centromere and telomere regions(Figure2),and chromo-some2had the highest InDel frequency and e-PCR validated marker density(Figure2;Table1).To confirm the efficiency of AGA-pipe,we compared the InDels/markers obtained from another two studies and our study.The number of e-PCR validated InDel markers(>6bp in2L€u et al.length)in this study was 12,944,which was 11times more than the reported 1,047in another study (Zeng et al.2013).We further compared the 12,944e -PCR validated markers in this study to the 986InDels developed by Zeng et al.(2013).The data we obtained from the paper of Zeng et al.(2013)only contained information of primers,making it hard to obtain the exact locations of InDels and therefore prohibited us from comparing all of these InDels against our own dataset.We filtered the 1,047pairs of PCR primers developed through MFEprimer2,and after removing mismatching primers and primers whose ampli fication products were not on target chromosomes,only 806InDels were accuratelylocatedFigure 1.Pipeline for InDels/markers development based on genome alignment and next-generation DNA sequencing(A )The strategy of InDel Markers Development Platform (IMDP)for InDel markers development.(B )Screenshot of IMDP interface in LONI pipeline environment.Reference genome:complete genome sequence,such as Nipponbare IRGSP build 5.Query:genome or a single DNA sequence,such as 93-11genome sequence.Nucmer :a Perl script pipeline in MUMmer 3for multiple closely related nucleotide sequences alignment.BWA :software package for NGS reads mapping.Pindel :software for detecting InDels.A and B in gray circles represent perl script used in InDels detection for different data sources.Set A and Set B are InDels from genome alignment and NGS mapping respectively.Primer3,a tool for PCR primers design.e -PCR,MFEprimer,which is used to evaluate the speci ficity of PCR primers.RIMD,Rice InDel Marker Database.Graphic pipeline platform for InDel markers development 3on rice chromosomes,among which 551were identi fied in our validated hq -InDel list.Within the 551hq -InDels,236overlapped with e -PCR markers identi fied by us,while the remaining 315markers in Zeng ’s study were filtered out because of lacking precise locations rather than our criteria.The number of md -InDels discovered using our InDel discovery pipeline was 90,548,which was 1.3times as many as that in a previous report (Shen et al.2004),the number of which was 68,073.Because of lack of detailed information of Shen ’s study,we could not do moredetailedFigure 2.Distribution pattern of hq -InDels and e -PCR validated markers of Nipponbare against 93-11The y-axis represents the physical distance along each chromosome in 100kb windows.The x-axis indicates the number of hq -InDels (in yellow)and corresponding e-PCR validated markers (in green).The gray areas show centromere regions.Table 1.Identi fication of InDels/markers between Nipponbare and 93-11using AGA-pipe Chromosome Nipponbaregenome (millionbase pairs)a 93-11genome (million base pairs)b md -InDels c hq -InDels d e -PCR validated markers Covered BAC (uncovered BAC)Frequency e Chr0145.047.212,3765,0201,745376(16) 4.4Chr0236.838.110,6362,2241,671340(19) 4.6Chr0337.341.89,0321,9961,400308(21) 4.2Chr0436.134.76,3962,136872217(78) 2.9Chr0530.131.26,4694,3101,015241(45) 3.5Chr0632.132.87,4943,9671,038251(30) 3.6Chr0730.427.97,3712,5031,110259(31) 3.8Chr0828.530.46,8102,800951254(24) 3.4Chr0923.921.76,4172,967954206(17) 4.2Chr1023.722.25,7852,777751182(21) 3.6Chr1131.223.05,6372,842668168(90) 2.6Chr1227.722.96,1252,524769196(73) 2.8Total382.8373.990,54836,06612,9442,998(465)3.7Nipponbare reference genome (IRGSP build5).93-11reference genome (GeneBank accession –GCA_000004655.2).md -InDels,the lengths of InDel polymorphisms were from 6to 100bp.d hq -InDels,md -InDels with no other polymorphisms more than 6bp in length existed in the 300bp region flanking the InDel.e Frequency,markers per BAC according to Nipponbare genome.4L €uet al.comparison between our markers with the reported 68,073InDels.We also used the same default settings to identify InDels between Nipponbare and Kasalath.Kasalath is a member of the aus group of O.sativa,which has been widely used in breeding and functional genomic studies for its specific traits such as early maturity and tolerance to drought and phosphate deficiency.The genome sequence of Kasalath was recently assembled(Sakai et al.2014).Overall,94,730md-InDels were detected using get-indels.pl program and38,846 hq-InDels were identified in Nipponbare/Kasalath genome (Table S3).The density of the identified hq-InDels was10.1per 100kb(Figure S4A).The lengths of about95%InDel poly-morphisms were6–50bp,which could be easily distinguished by gel electrophoresis(Figure S4B,C).Development of InDel markers based on NGS-pipeGiven that molecular markers are cultivar-specificity,tools for developing markers of any two specified cultivars will be of much significance in basic research and breeding.We used NGS-pipe for InDel markers development based on NGS mapping and tested it using two landrace cultivars widely planted in China,Wu-Yun-Geng7(WYG7,an Oryza sativa japonica cultivar)and Guang-Lu-Ai4(GLA4,an Oryza sativa indica cultivar).We re-sequenced WYG7and the average re-sequencing depth was approximately35Âusing Illumina Genome Analyzer II platform.The NGS data of GLA4was retrieved from National Center for Biotechnology Information(NCBI) (See methods).Since re-sequencing depth affects the cover-age of a target genome and is one main evaluation criterion for NGS data quality(Sims et al.2014),wefirst checked the correlation between theoretical sequencing depth and genome coverage.It was no surprise tofind that theoretical sequencing depth and genome coverage were positively correlated(Figure3A).Regression analysis further suggested that the relationship between genome coverage and theoretical depth might meet reciprocal function y¼(aþbx)/x(a<0,b>0,x and y represent theoretical depth and coverage breadth,respectively,Figure3B).Thereciprocal Figure3.The correlation analysis between genome coverage,InDels number and theoretical sequencing depth(A)The positive correlation between genome coverage and sequencing depth of WYG7against Nipponbare reference genome. Blue dotted line shows that the coverage of WYG7NGS data in the reference genome is94.39%,close to the largest theoretical coverage(97.22%)when the sequencing depth is10.(B)The regression analysis shows that breadth of coverage(y)and theoretical depth(x)might meet the reciprocal function:y¼(aþbx)/x.(C)The total number of InDels among WYG7,GLA4and Nipponbare,indicating that the amount of InDels is positively correlated to the sequencing depth.The proportion of md-InDels is 12%to20%.(D)The distribution pattern of InDels and e-PCR validated markers in WYG7/GLA4genome when sequencing depth is 10,showing that the lengths of98.6%md-InDel markers in WYG7/GLA4are6–50bp.Graphic pipeline platform for InDel markers development5function we obtained is y¼(–0.5139þ0.9722x)/x,in which0.9722represents the largest theoretical coverage, and R2represents the correlation coefficient,respectively (Figure3B).When the sequencing depth was10,the NGS data coverage of WYG7in the reference genome was94.39% (Figure3A),close to the largest theoretical coverage(97.22%), suggesting that a cost-effective sequencing depth might be10Â.We also checked the correlation between InDels number and re-sequencing depth of WYG7and GLA4.Similar to a previous report(Arai-Kichise et al.2011),the number of InDels was positively correlated with re-sequencing depth (Figure3C).However,the percentage of md-InDels was relatively stable,ranging from12%to20%regardless of the re-sequencing depth(Figure3C;Table S4).It was not surprising tofind that InDels number of japonica/japonica cultivars (such as WYG7/Nipponbare)was much less than that of japonica/indica cultivars(such as WYG7/GLA4and Nipponbare/GLA4)(Figure3C).The number of md-InDels was34,794in WYG7/GLA4genome,33,432in GLA4/Nippon-bare genome,and6,301in WYG7/Nipponbare genome,all adequate for further e-PCR validation(Table S4).Therefore, we developed md-InDel markers in WYG7/GLA4genome with re-sequencing depth of10.Consistent with the distribution pattern of InDels in Nipponbare/93-11genome,the telomere and centromere regions of chromosomes could not be covered by md-InDels and e-PCR validated InDels in WYG7/ GLA4genome either(Figure S5).The length distribution of NGS-pipe results are similar to the pattern of AGA-pipe results,in which98.6%md-InDel markers in WYG7/GLA4were 6-50bp in length(Figure3D;Table S5).These results suggested that the NGS-pipe is efficient for InDel marker development.Development of InDel markers between japonica and indica subgroup rice cultivarsThe application of NGS technologies is a great step towards fast revealing of the genomic diversities of rice germplasms (Kilian and Graner2012),as well as facilitating the develop-ment of japonica/indica subgroup markers in rice.To develop genome-wide subgroup markers between japonica and indica cultivars,we collected NGS data of13rice cultivars in addition to WYG7,GLA4,Nipponbare,and93-11(Table S6).To avoid selecting closely related cultivars during rice domestication and breeding,we only chose the ones from diverse geo-graphic distributions in Asia or America(Table S6).All of these NGS data werefirstly processed by Trimmomatic as described above(Table S7),then compared individually with the Nipponbare reference genome sequence.The number of md-InDels detected in these cultivars varied from1,641(NK-58/ Nipponbare)to56,867(IR36/Nipponbare),demonstrating a different number and density of InDels on each chromosome between japonica and indica subgroups(Figure4A;Table S7). Totally,133,420md-InDels were obtained in all15cultivars against the Nipponbare reference genome sequence.Among them,18,309md-InDels were also found in93-11/Nipponbare (Figure S5,Table S8).It was interesting that the InDel densities of japonica/Nipponbare genomes varied significantly on all chromosomes,while the InDel densities of indica/Nipponbare genomes was very close to each other(Figure4A).The18,309 InDels were further analyzed by hierarchical cluster analysis.As expected,japonica cultivars and indica cultivars were clustered on two different branches(Figure4B).Furthermore,we identified2,681japonica/indica subgroup md-InDels,out of which1,042passed designing primers and e-PCR criteria(blue bar in Figure4C).We randomly chose95 subgroup markers to perform PCR and denatured PAGE followed by silver staining in12cultivars(see Table S6).The verification results showed that two markers($2.1%)gen-erated no polymorphism and seven markers($7.4%)gen-erated polymorphisms within japonica or indica subgroup cultivars,which suggested the accuracy of our developed japonica/indica subgroup InDel markers reached90.5% (Table S9;Figure S7).These subgroup specific markers can be used to develop key markers for rice gene mapping and molecular breeding.As mentioned above,one of the preconditions for molecular breeding is functional genomics analysis.The correlation between japonica/indica subgroup InDel markers in this study and151reported agricultural trait genes was also investigated(Table S10),such as yield,grain quality,resistance to bacteria and pest,plant height and tillering.Grain size, grain number and grain weight are three important traits contributing to grain productivity.In our analysis,25grain production quantitative trait loci(QTLs)were linked with our japonica/indica subgroup markers(Table S10),such as Ghd8/ DTH8(a QTL conferring pleiotropic effects on grain productivity,plant height,and heading date)(Yan et al. 2011),Ghd7(a QTL regulating growth,development and stress response)(Weng et al.2014),FLO2(a QTL involved in regulation of rice grain size and starch quality)(She et al. 2010),GS5(a QTL regulating grain size and yield)(Li et al.2011) (Table S11).All of those agricultural trait QTLs are very important in rice productivity,whose verification during MAS could be performed by our japonica/indica subgroup markers. In total,the japonica/indica subgroup markers appeared within one mega base of89trait genes downstream or upstream (about4cM,Table S11),which might suggest that these markers could be used for testing the89agricultural trait genes in molecular breeding.A web-based database for rice InDel markersAll of the205,659md-InDels(133,420md-InDels from NGS data against Nipponbare,90,548md-InDels from93-11against Nipponbare,Figure S6)and12,944e-PCR validated InDel markers developed in this study were deposited into a user-friendly database,RIMD(http://202.120.45.71/),which is a website developed using LAMP(Linux,Apache,MySQL and PHP)and openly accessed for researchers to browse and query rice InDels/markers(Figure5).To further facilitate the application of these InDels/markers in molecular breeding,we imported5,589agronomically important genes from China Rice Data Center(/gene/),Oryzabase (http://www.shigen.nig.ac.jp/rice/oryzabase/gene/list) (Yamazaki et al.2010)and PMRD(/)(Cui et al.2012).These genes were classified by trait ontology (Jaiswal et al.2002),including anatomy and morphology trait, biochemical trait,growth and development trait,quality trait, stature or fertility trait,stress trait,and yield trait.The RIMD database hasfive functional modules(Figure5):‘Jap/In markers’for key markers of japonica/indica subgroup cultivars;‘Search’for InDels/Markers by ID,gene locus,6L€u et al.BAC/PAC or chromosome region and the details of classical markers were shown in ‘Search result ’(Figure S8);‘Trait Genes ’for agronomically important genes (Figure S9);‘Gbrowse ’for Generic Genome browser (Stein et al.2002);‘Primer3plus ’for primer design online,‘Tools and Download ’for useful scripts,datasets developed in this research,and useful links of rice research websites in ‘More ’;‘Help ’for tutorial of RIMD and frequently asked questions.Therefore,RIMD provides a friendly web interface that allows users to easily search e-PCR validated InDels for rice MAS or other basic research,such as map-based gene cloning.RIMD also provides the pipeline and scripts developed in this study for other researchers to develop InDels in other rice cultivars or develop markers in other plants with reference genome sequences.For users who intend to obtain InDels coordinates on the latest rice reference genome (IRGSP7)(Sakai et al.2013),we provided a converted General Feature File (GFF)which can be downloaded on RIMD (http://202.120.45.71/attachments/article/10/indels_gff_to_msu7.7z ).DISCUSSIONTo increase the production of high-quality food to cope with global environmental change and population increase,marker-assisted selection (MAS)could be one of most promising and effective strategies (Collard and Mackill 2008).However,the biggest limitation of MAS in plant breeding is the cost of marker development and utilization (Collard et al.2005;Collard and Mackill 2008).High-throughput genotyping platforms based on DNA sequencing or DNA microarray have been rapidly developed in the past few years (Chen et al.2014;Yu et al.2014).These techniques generally require special equipment and complicated bio-informatics analysis.With the advantage of high density of InDels in genome and its direct application based on PCR rather than sequencing,we focused on moderate length InDel markers development for ordinary laboratory and breeding organization.Instead of traditionally time-consuming and labor-intensive approaches of developing InDel markers,which needs to employ Basic Local Alignment SearchToolFigure 4.Characterization of md -InDels and distribution of japonica /indica subgroup InDel markersThe overall number of md -InDels in indica cultivars against Nipponbare genome was more than that in japonica cultivars against Nipponbare genome.(A )The InDels density (per mega bases)of each cultivar against Nipponbare genome on each chromosome.(B )The hierarchical cluster analysis of the 18309InDels in 17cultivars genome.japonica cultivars:Canella De Ferro (CDF),Chodongji (CDJ),Haginomae Mochi (HM),Lemont (LEM),Nongken-58(NK),Omachi (Om),Wuyungeng-7(WYG)and Nipponbare (NIP).indica cultivars:Popot 165(P165),Ai-Chiao-Hong (ACH),Gie 57(G57),Guang-lu-ai 4(GL4),Guan-Yin-Tsan (GYT),IR36,JC91,93-11and Leung Pratew (LP).(C )2681md -InDels were identi fied as japonica /indica subgroup InDel markers,which distribute unevenly on 12chromosomes and are shown in blue segments (e-PCR veri fied markers)and red dots (PCR validated markers).Graphic pipeline platform for InDel markers development 7。