Fine mapping QTLs using Recombinant-Inbred HS and In-Vitro HS William Valdar Jonathan Flint, Richard Mott Wellcome Trust Centre for Human Genetics
Heterogeneous Stocks Pseudo-random mating for N generations typical chromosome pair 8 inbred lines eg, N=30: 3.4cM (=100/30) average distance between recombinants
Cost of mapping with HS Need to genotype markers at very high density (sub centimorgan) Expensive to genotype whole genome (eg 3000 markers for 30 generation HS) How can we reduce genotyping cost ? Use multiple phenotypes (value for money) Two genetic strategies: RIHS Recombinant Inbred Heterogeneous Stock IVHS In vitro Heterogeneous Stock
Recombinant Inbred HS (RIHS) X 20 generations HS RIHS
Recombinant Inbred HS (RIHS) X 20 generations HS RIHS Genotype each RIHS line once Keep stock, eg, as embryos Distribute RIHS lines to labs for phenotyping
Recombinant Inbred HS (RIHS) X 20 generations HS RIHS Advantage over standard RI : resolution Advantage over standard HS: cost Genotype each RIHS line once Keep stock, eg, as embryos Distribute RIHS lines to labs for phenotyping
RIHS for mapping modifier QTL X 20 generations X HS RIHS inbred F1 (may contain knockout or transgene) modifier search
How many RIHS do we need for effective fine- mapping? Are there other HS strategies to reduce genotyping…?
In Vitro HS (IVHS) HS donor recombinant HS sperm F1 IVF Fertilize inbred dam with HS sperm meiosis
IVHS-1 genotype donors at high resolution HS donor recombinant HS sperm F1 IVF meiosis
IVHS-1 genotype donors at high resolution HS donor recombinant HS sperm F1 IVF pass 1 pass 2 F1 markers meiosis
IVHS-2 HS donor recombinant HS sperm F1 IVF treat as average of donor chromosomes no further genotyping meiosis genotype donors at high resolution
Simulations Compare strategies RIHS, IVHS-1, IVHS-2 by simulation
Simulations Compare strategies RIHS, IVHS-1, IVHS-2 by simulation Simulate 25cM chromosome with single additive QTL placed randomly
Simulations Compare strategies RIHS, IVHS-1, IVHS-2 by simulation Simulate 25cM chromosome with single additive QTL placed randomly Type 100 SNP markers
Simulations Compare strategies RIHS, IVHS-1, IVHS-2 by simulation Simulate 25cM chromosome with single additive QTL placed randomly Type 100 SNP markers 30 generation HS
Simulations Compare strategies RIHS, IVHS-1, IVHS-2 by simulation Simulate 25cM chromosome with single additive QTL placed randomly Type 100 SNP markers 30 generation HS Vary –QTL effect size (1% to 50%) –# RIHS lines used (40, 80, 120) –Sample size (400 to 2000 total number of pups)
Simulations Compare strategies RIHS, IVHS-1, IVHS-2 by simulation Simulate 25cM chromosome with single additive QTL placed randomly Type 100 SNP markers 30 generation HS Vary –QTL effect size (1% to 50%) –# RIHS lines used (40, 80, 120) –Sample size (400 to 2000 total number of pups) Also investigate for IVHS-1 –Marker density –SNPs v Microsatellites –# HS generations
Evaluating the simulations Evaluation –Perform 1000 simulations per condition –Analysis performed with HAPPY –Probability of detecting a QTL (must be a marker interval with adjusted HAPPY Pvalue < 1%) –Mapping accuracy
Detecting a significant locus Pass rate = % times most significant marker interval has (corrected) P-value less than 0.01
Detecting a significant locus Pass rate = % times most significant marker interval has a corrected P-value less than 0.01 consistent across population sizes 5%
Mapping accuracy for significant loci Mean mapping error = average distance between true QTL and the predicted locus mapping error (cM) predicted QTLtrue QTL
Mapping accuracy for significant loci Mean mapping error = average distance between true QTL and the predicted locus mapping error (cM) predicted QTLtrue QTL
Varying marker density and marker type IVHS-1 strategy with 5%QTL, 1200 pups Vary number of markers over a 3cM region
Varying marker density and marker type IVHS-1 strategy with 5%QTL, 1200 pups Vary number of markers over a 3cM region Microsats betterMicrosats = SNPs ~0.05cM
Varying number of HS generations IVHS-1 strategy with 5%QTL, 1200 pups
Varying number of HS generations IVHS-1 strategy with 5%QTL, 1200 pups optimum [5,15]
Conclusions RIHS and IVHS strategies: low genotyping cost without sacrificing mapping resolution IVHS is short term mapping strategy RIHS takes longer, costs more but is long term strategy of choice. 100 RIHS lines is sufficient for mapping isolated additive QTLs but may not be enough for multiple QTLs identifying epistatic effects Suitable HS: need only 15 generations Paper submitted to Mammalian Genome (preprints available)