A family-based imputation algorithm

Transcription

1 EAAP th Annual Meeting of the European Federation of Animal Science Belfast UK, 9 Aug Sept 06 A family-based imputation algorithm for high- and lowcoverage sequence data Mara Battagin* Gregor Gorjanc Serap Gonen Roberto Antolin John Hickey * mara.battagin@roslin.ed.ac.uk

2 Introduction Genomic Selection.0 Sequence pigs and chicken animals at high- and lowcoverage Phasing and imputation for all individuals > 00,000 pigs and 0,000 chicken with sequence information Who? Distribution of sequencing resources Genomic Selection Unequal Key sires Equal i.e. All population at low coverage Focal individuals Individuals whose haplotypes are shared with many other individuals in a population. AlphaSeqOpt (by Gonen et al.)

3 Introduction heuristic phasing algorithm for high- and low-coverage sequence data AlphaSuite program Family-based method Developed for sequence data PGS PGM MGS MGM Dam Focal individual

4 Method Reads nref nalt Snp A family-based imputation algorithm using high- and lowcoverage sequence data

5 Method Reads Snp nref nalt Variant Calling % error rate Genotype threshold >0.99 Iterate until convergence Build the trio consensus (grandparent parent progeny) A family-based imputation algorithm using high- and lowcoverage sequence data Founder Assignment Phasing based on parents informative markers Phasing based on parent-progeny informative markers

6 Method Reads Snp nref nalt Genotype Variant Calling % error rate Genotype threshold >0.99 Iterate until convergence Build the trio consensus (grandparent parent progeny) A family-based imputation algorithm using high- and lowcoverage sequence data Founder Assignment Phasing based on parents informative markers Phasing based on parent-progeny informative markers

7 Method Reads Snp nref nalt Genotype Variant Calling % error rate Genotype threshold >0.99 Iterate until convergence Build the trio consensus (grandparent parent progeny) A family-based imputation algorithm using high- and lowcoverage sequence data Founder Assignment Phasing based on parents informative markers Phasing based on parent-progeny informative markers PGS PGD MGS MGD Focal Id Dam

8 Method Reads Snp nref nalt Genotype Variant Calling % error rate Genotype threshold >0.99 Iterate until convergence Build the trio consensus (grandparent parent progeny) A family-based imputation algorithm using high- and lowcoverage sequence data Founder Assignment Phasing based on parents informative markers Phasing based on parent-progeny informative markers PGS PGD MGS MGD Focal Id Dam

9 Method Reads Snp nref nalt Genotype Variant Calling % error rate Genotype threshold >0.99 Iterate until convergence Build the trio consensus (grandparent parent progeny) PGS PGD MGS MGD Focal Id Dam A family-based imputation algorithm using high- and lowcoverage sequence data Founder Assignment Phasing based on parents informative markers Phasing based on parent-progeny informative markers PGS PGD MGS MGD Focal Id Dam

10 Method Algorithm tested through Simulation Simulate true genotypes (AlphaSim) 0 chromosomes Morgan length 00,000 segregating sites scenarios Within Family imputation Selected families as replicates Multiple Families imputation External cattle pedigree (,8 ids) Simulate sequence data different scenarios Number of family members sequenced (from to 7) Sequencing depth (from 0 to 0x) Extracted generations for 00 sires Total,6 animals Accuracy of phasing the focal individual for 8, scenarios Key sires Budget 0,000 Focal individuals AlphaSeqOpt (by Gonen et al.) Equal distribution

11 Results Accuracy of phasing within family Cost of the library/individual = 0 Cost of x whole genome = 80

12 Results Accuracy of phasing multiple families Percentage of Correct and Wrong phased gamete by read depth Key s Min 0.00 Mean. Max 99.0 All population Min. Mean 9.0 Max Focal individuals Min 0.00 Mean.8 Max 99.

13 Conclusions Heuristic phasing algorithm for high- and low-coverage sequence data Able to increase the sequence information in livestock population Good accuracy of can be achieved ToDoList: Test it with the real data!

14 Acknowledgements Thank you! John Hickey Gregor Gorjanc Roberto Antolin Serap Gonen Diarmaid De Burca Andrew Derrington Chris Gaynor Paolo Gottardo Janez Jenko Daniel Money Roger Ros Freixedes Maria Sanchez Perez Masoud Shirali Adriana Somavilla

15 Method multiple-families sequence data Coverage Unequal distribution of resources Key s Focal Individuals Equal distribution of resources 0x 96 6 x 0 x.7x,6 x 0 0x x 09 0x Cost ( ) 0,000 9,60 9,60 Cost of the library/individual = 0 Cost of x whole genome = 80

16 Method Variant calling Reads nref Snp nalt Genotype 0 0 PatHap MatHap Iterate until convergence Parameters: Error Rate = 0.0 Genotype threshold = 0.999

17 Method Phasing based on parents informative markers Phase the markers inherited from homozygous parents PatHap 0 0 MatHap 0 0 Progeny PatHap MatHap 0 Genotype 0 Iterate until convergence

18 Method Phasing based on parent-progeny informative markers Phase the heterozygous markers of individuals that have a parent and progeny with opposite alleles Grand Dam 0 0 Progeny Iterate until convergence 0 0

19 Method Founder Assignment Identify trios (grandparent, parent, individual) by assigning grandparents to individuals using the heterozygous markers of their parents PGS PGM MGS MGM Dam Progeny Progeny PGS PGD PGD PGD Iterate until convergence

20 Method Founder Assignment Identify trios (grandparent, parent, individual) by assigning grandparents to individuals using the heterozygous markers of their parents PGS PGM MGS MGM Dam Progeny Progeny Forward PGS PGD PGD PGD Iterate until convergence

21 Method Founder Assignment Identify trios (grandparent, parent, individual) by assigning grandparents to individuals using the heterozygous markers of their parents PGS PGM MGS MGM Dam Progeny Progeny Forward PGS PGD PGD PGD Backward

22 Method Founder Assignment Identify trios (grandparent, parent, individual) by assigning grandparents to individuals using the heterozygous markers of their parents PGS PGM MGS MGM Dam Progeny Progeny Forward PGS PGD PGD PGD Backward Use only congruent chunks

23 Method Build the trio consensus Identify chunks of haplotypes within trios and use the chunks to impute missing markers between members of the trio. GrandDam Progeny Iterate until convergence

24 Individuals Results Accuracy of phasing multiple families Chunk of a paternal gamete (00 SNP) for,6 individuals sequenced according to three different strategies Unequal distribution of sequencing resources Key s Focal Individuals Equal distribution of sequencing resources Snp Snp Snp Yield: Correct Wrong Missing