With the development of resequencing and genotype imputation methods large amounts of SNPs are available. Many of the SNPs are thought to be redundant. TSR was developed mainly to select appropriate SNPs to improve the genome prediction and both genotype and dosage are supported. Several methods can be chosen: minimal allele frequency, linkage disequilibrium (similar to plink prune method, but for TSR the slip window is based on physical position, not SNP counts), cluster (quasi nearest neighbour), average distance and p-value from genome wide association studies. The program also has other functions. It can be used to estimate the linkage disequilibrium decay and it's fast and memory efficient. It can also be used to estimate the variance of SNP effects and build weighted genome relationship.
The program was written by C++ and compiled using gcc 4.4.7 on a 64-bit Linux version. Current version: TSR_float_v0.1.0 (genotypes are stored in floats and compatible with software MGP_float (https://github.com/codeatcg/MGP)). For current version when the number of chromosomes is more than 1 the output positions of pruned SNPs may be not correct. A new version fixed the bug and it has new features. The new version supports genotype in bed format (a genotye takes 2 bits) and supplies another two methods (sequential core and tuned sequential core) to select SNPs. The new version will be released soon.
vcf2tag TSR vcf2tag [options]
Convert genotypes in VCF format to binary format. Binary genotypes are used to build genome relationship and get appropriate SNPs to improve the genome prediction and genome wide association.
//Convert Vcf to Binary Genotype:
--vcf <FILE> input vcf file
--gType <STR> **genotype or dosage
--outGen <FILE> output binary genotype
--outPos <FILE> **output coordinate file
//Prune SNPs:
--inHeader <FILE> **vcf file or vcf header file
--inGen <FILE> **input binary genotype
--inPos <FILE> input coordinate file
--outGen <FILE> output binary genotype
--outPos <FILE> **output coordinate file
--win <INT> window size, by default:50000
--overlap <INT> overlap size, by default:5000
--value <FLOAT> relation between SNPs,by default:0.9
--verbose output detailed information for each window
--cluster SNPs pruned by cluster
--ldprune SNPs will be removed if they are in tight linkage with any others, similar to plink
--unif SNPs pruned by average distance
--mafwin SNPs pruned by minimal allele frequency in a window
//Build Relationship:
--inHeader <FILE> **vcf file or vcf header file
--inGen <FILE> **input binary genotype
--gType <STR> **genotype or dosage
--addMethod <INT> 1 (formula according to Vanraden) or 2 (similar to GCTA)
--bin2kin build genome relationship
--outKin <FILE> output binary kinship
--outTxt <FILE> output kinship in text
--part genotype file is partial and the output kinship file need to be merged
ld TSR ld [options]
Linkage disequilibrium decay was estimated by the correlation coefficients between markers, which was thought to be comparable with the method based on linkage phase. All SNP pairs are evaluated, but it’s optional that just using part of pairs if maker densities are high.
--pos <FILE> position file
--genotype <FILE> input binary genotype file
--inHeader <FILE> vcf file or vcf header file
--out <FILE> output file
--win <INT> window size, by default: 500000
--thin chromosome segments are independent, without this parameter all SNP pairs with distance less than the window will be considered
wkin TSR wkin [options]
Estimated breeding value (EBV), relationship and binary genotype are needed to build weighted genome relationship. It’s similar to iteration ssGBLUP but the analysis can be parallelized by chromosomes or chromosome segments.
--inKin <FILE> input binary kinship
--inEbv <FILE> input ebv file
--inGeno <FILE> input binary genotype file
--posFile <FILE> input position file (match the binary genotype file)
--outKbin <FILE> output weighted binary kinship
--outKtxt <FILE> output weighted kinship in text
--varSnp <FILE> output variance of SNPs effects
--varWin <FILE> output genetic variance explained by SNPs in a window
--winMet <INT> method to calculate variance explained by SNPs in a window (1,2,3), by default: 1, method 1- similar to ssGWAS, method 2- sum the effect variances of SNPs in a window, method 3- similar to method 2 but using sliding window
--winSize <INT> number of SNPs in a window, by default: 20
--overlap <INT> number of SNPs in an overlap region, by default: 5
--varLimit <FLOAT> threshold to remove SNPs with small effect variances, by default: 0
--part genotype file is partial and the output kinship file needs to be merged
qtn TSR qtn [options]
Trait specific markers were selected by p value from genome wide association studies.
--pFile <FILE> p-value file
--pvalue <FLOAT> threshold to select trait specific SNPs,range (0,1],by default 0.01
--allGen <FILE> input binary genotype (all genotype)
--allPos <FILE> input coordinate file, all locis
--qtnGen <FILE> output triat specific SNPs (binary genotype)
--qtnPos <FILE> output positions of trait specific SNPs
--cor <FLOAT> threshold to select independent SNPs,range [0,1],by default 0.9
--pruneGen <FILE> input binary genotype (pruned genotype)
--prunePos <FILE> input coordinate file, pruned locis
--indepGen <FILE> output independent SNPs (binary genotype)
--indepPos <FILE> output positions of independent SNPs
--sel <CHAR> select QTNs and locis independent of them, values can be 'qtn', 'indep' or 'both', by default 'both'
--rSnp <INT> number of SNPs kept in memory at a time,by default 1000
--inHeader <FILE> vcf file or vcf header file
EBV, p value, kinship and genotype formats are compatible with software MGP_float and the output files can be used by TSR, directly.
If you find bugs please email me.
Email: miaozepu@genomics.cn
Institute: Applied agriculture of BGI
Free for research. For commercial use please contact P_agro_pmo@genomics.cn