gwasurvivr

gwasurvivr performs genome-wide survival analysis of single nucleotide polymorphisms (SNPs) to identify associations between genetic variants and time-to-event (survival) outcomes using Cox proportional hazards models in R/Bioconductor.

Key Features:

• Input format support: Accepts VCF files from Michigan or Sanger imputation servers, IMPUTE2 files, and PLINK formats for genotype/imputation data.
• Scalability: Handles large-scale genome-wide datasets, including analyses involving millions of SNPs.
• Optimized Cox proportional hazards implementation: Modifies the R package 'survival' by first fitting covariates without the SNP and using those estimates as starting values to reduce iterations for SNP parameter estimation.
• Benchmarking and performance: Has been benchmarked against genipe, SurvivalGWAS_SV, and GWASTools and demonstrated faster runtimes and improved scalability with increasing sample size, SNP count, and covariate number.

Scientific Applications:

• GWAS with time-to-event data: Detects associations between SNPs and survival outcomes in genome-wide association studies incorporating time-to-event phenotypes.
• Genetic epidemiology of disease progression: Investigates how genetic variants influence disease progression or patient survival in large cohorts.

Methodology:

Parses VCF (Michigan/Sanger), IMPUTE2, and PLINK inputs and applies a Cox proportional hazards framework with a modified estimation strategy that fits covariates without the SNP to obtain starting values for SNP-specific parameter estimation.