Shared Genomic Segments (SGS) Analysis Method: Application to Extended Utah Pedigrees at High Risk for ASD

Background:  Shared Genome Segments (SGS) analysis is a method used to identify shared regions of DNA among cases using high-density SNP genotypes. These regions can then be used to prioritize subsequent experiments to search for causal sequence variants. This approach will be particularly advantageous when functionally relevant variants are regulatory, allowing an investigator to focus on full targeted sequencing with an SGS region, a compliment to the whole exome approach.

Objectives:  The method will be described in detail, including a flow chart of the steps required to apply the method successfully, input file formats, examples of run commands, example output, and a visualization tool for output.  We have analyzed extended pedigrees at high risk for Autism Spectrum Disorder (ASD). Different options to apply the method using this data resource will be demonstrated.

Methods:  The SGS test identifies if the length of consecutively shared SNPs, identified as identical-by-state, or IBS, is longer than expected by chance. IBS is established by determining if allelic types at sequential SNPs are consistent across cases. IBS does not infer identity-by-descent (IBD; the same inherited segment from a common ancestor) which is our true interest. However, if the length of SGS shared IBS is significantly longer than by chance, given the known relationships, then IBD is implied. Theoretically, chance IBD sharing in distant relatives is extremely improbable. Genomewide statistical significance can be found for pedigrees with at least 15 meioses. SGS is efficient, and can have more power than traditional association tests.

Results:  Pedigree members were genotyped using the Illumina HumanOmniExpress12v1.1 array.  While SGS can be used to test for sharing across all pedigree cases, due to heterogeneity we did not expect all affected cases to share the same segments in these large pedigrees. We therefore started with a program option, weighted pairwise SGS (wpSGS), which combines sharing across all possible pairs, weighted by number of meioses between the pairs. We show results of wpSGS; with the increased length illustrated by the ratio against both internal pedigree controls, and also to unrelated controls (pairs selected from 168 North-West European individuals—CEU/GBR—from the 1000 Genome Project). Statistical significance for observed SGS regions (p-values) are estimated by generating null genome simulations (a user-specified number), based on a genetic model estimated from the 1000 Genome Project controls. 

Conclusions:  We present a flow chart and example commands to use the software with different analysis options and control sets. We show input file structures and format of the output. We show how the null genomes are simulated; including incorporation of linkage disequilibrium (LD) and a recombination model. We show how SGS can also be set up for validation testing of sharing across subsets of pedigree cases identified by wpSGS. We demonstrate additional software to visualize and prioritize results. Through this demonstration, we show that SGS is an easily applied, efficient, useful pedigree analysis tool in the search for susceptibility variants for complex diseases, such as ASD.