Additional file 1 of Constructing germline research cohorts from the discarded reads of clinical tumor sequences

Additional file 1: Figure S1. Histogram of broad cancer types in the full tumor cohort. Figure S2. Coverage histogram. Figure S3. Imputation accuracy by filtering criteria. Figure S4. Robustness of imputation correlation estimate. Figure S5. Copy neutral loss of heterozygosity calling (CN-LOH). Figure S6. Distribution of imputation correlation across all (pre-filtered) HapMap3 variants by imputation scheme (x-axis and color code). Figure S7. Distribution of imputation correlation by INFO score and coverage. Figure S8. Cumulative imputation accuracy. Figure S9. Imputation correlation by variant type. Figure S10. Imputation correlation for pseudo-SNP indels. Figure S11. Manhattan plot of imputation correlation across panel versions. Figure S12. Distribution of imputation allelic error across sequencing panels. Figure S13. Distribution of imputation allelic error by coverage and panel. Figure S14. Variance in imputation error explained by technical features. Figure S15. Imputation error by tumor TMB and FFPE sample. Figure S16. Imputation error by tumor purity. Figure S17. Percent of SNPs with high levels of error at somatically altered regions. Figure S18. HLA homozygosity calling accuracy. Figure S19. PRS imputation accuracy. Figure S20. Breast PRS error. Figure S21. PRS mean error by panel. Table S1. Association of somatic features with imputation error. Table S2. Number of somatic SNVs per sample that overlap a common reference panel variant in PCAWG tumor WGS data. Table S3. EGFR associations with race and ancestry.