Evaluaciones genómicas utilizando secuencias del proyecto de los 1000 genomas bovinos

Abstract

[EN] regression approach for genetic evaluation using population sample of 450 Holstein animals , with full - sequence data from the 1000 bull genomes project is p resented in this thesis . This approach is based on the assumption that haplotypes from sequencing data are in strong er li nkage disequilibrium (LD) with Quantitative Trait L oci (QTL) than marker s from SNP chips . This study focuses on the extraction of haplo types in the population and their incorporation in the whole sequence prediction model . In total, 38,319, 258 SNPs (and indels) from Next G eneration Sequencing ( NGS ) were included. V ariants with Minor Allele Frequency ( MAF< 0.025 ) were discarded leading a t otal of 13 , 912 , 326 SNPs available for the analyses . H aplotypes were obtained from version 3 of findfhap.f90 software . The number of SNPs in the haploblock s ranged from 79 9 in BTA 13 to 1285 in BTA 12 , with a mean of 924 SNP ( 166 , 552 pb). The h aplotypes wit h a frequency below 1% were around 97% in all chromosomes . These haplotypes were ignored leaving 1 53 , 428 haplotypes for subsequent analys e s . Each haplotype was identified by chromosome and segment where it is located as well as the ordered number of the ha plotype within the segment. The haploblocks were then used to predict four economically important traits: kg of protein (Prot) , Global Type Index (IGT), Somatic Cell Score ( S C S ) and Days Open (DO). The phenotypic values were the MACE proofs provided by the Spanish Holstein Association CONAFE. The phenotypic data were merged with the haplotype file and a Bayesian model was implemented to predict G enomic E stimated B reeding V alue (GEBV) . E stimated h aplotypes had a large contribu tion to the total variance of GEBV ( between 3 2 and 99.9 % ). Most of the haplotypes for Prot, IGT and DO have low - intermediate frequencies while haplotypes found for SCS are mostly at low frequencies. We expect that these haplotypes will give us additional information to SNP genotypes on those less common variants , as SNP b eadchips are designed to genotype intermediate - high MAF . In order to reduce the number of haploblocks neede d to perform genomic prediction , a subsets of haploblocks that contain ed haplotypes with large effects were selected. A total of 1264 haplotype s exce eded the genome wide threshold of 3 standard deviation (sd) above mean (in absolute value) for Prot, 1909 for IGT, 851 for SCS and 1450 for DO distributed along the genome. In the second criterion , those with effect estimate (in absolute value) larger than 1 s d above mean which led to a total of 44 , 319 haplotype s f or 20 Prot, 39 , 975 for IGT, 46 , 132 for SCS and 42 , 878 for DO . Then, haplotypes selected in each criteria were subjected to a new analysis. The proportion of the genetic variance esti mated values corresponding to the haplotypes effect was 1.06, 5.24, 15.29 and 11.64% for Prot, IGT, SCS and DO , respectively , using the first criterion ( 3 sd ) and 10.92, 33.30 and 53.93 % for Prot, SCS and DO , respectively , using the second criterion (1 sd ) . Genomic predictions using only a set of app ropriately selected haploblocks can provide additional information to GEBV prediction , and should be considered in more in - depth

[ES] El objetivo de este trabajo sería el desarrollo de estrategias que incorporen información de secuencias en las evaluaciones genéticas. Para ello se utilizarán las secuencias del consorcio de los 1000 genomas bovinas. Se evaluará la capacidad predictiva de esta información y se comparará con la obtenida utilizando chips de SNPs de alta densidad. Durante el periodo de investigación se prevé que el alumno/a desarrolle además conocimientos de programación en fortran, R y/o java.