Fig. 2: Germline genetic variants underlie most putative ancestry-associated dependencies.

a The association between all genetic variants (from SNP6 genotyping arrays) and all putative ancestry-associated genetic dependencies was computed by linear regression with correction for cancer lineage as a covariate. Nominal p-values were adjusted for multiple hypothesis testing with the Benjamini–Yekutieli⁴³ procedure. Genes with at least one significant association (d-QTL) are indicated in red (n = 33). Raw data are described in Source Data 2AB. b The minor allele frequency of the most significant variant from a is indicated for the ancestry group with the lowest and highest frequency in the cell line collection. Cell lines of predominantly American or South Asian ancestry were not included in this analysis because the sample sizes for these ancestry groups are too low. Raw data are described in Source Data 2AB. c Position of the most significant germline variant from a relative to the Transcription Start Site (TSS) of the dependency gene in question. Variants that are within 1 megabase of the TSS are indicated in red (n = 29) and those that are greater than 1 megabase or are on a different chromosome are black (n = 20). Raw data are described in Source Data 2C. d Association between each d-QTL Single Nucleotide Variant (SNV) and expression of the dependency gene in question. Genes in this analysis were filtered to include only those with at least one significant d-QTL association (indicated in red in a). CDKN2B was excluded from this analysis because the association between the d-QTL variant and CDKN2B expression resulted from a technical artifact (Supplementary Fig. 2). Raw data are described in Source Data 2D. e Median expression across cell lines for all genes (top), all genes classified as common essential in The Cancer Dependency Map 22Q1 release (middle), and all ancestry-associated genetic dependencies (bottom). Raw data are described in Source Data 2E.