Ploidy-specific lethal (PSL) score analysis

Genes were assigned a binary classification (essential or non-essential) based on cutoffs established by Project Achilles. In the database, a score of −1 is assigned to a gene when its depletion in a given cell line results in a viability defect equal to the depletion of a curated list of gold standard common-essential genes^49–50. Based on this scoring system, we defined any gene with a score ≤ −1 for a given cell line as essential. We then compared the fraction of cell lines in the WGD^– and WGD⁺ groups where a gene was essential. When a gene was essential in a significantly greater fraction of WGD⁺ cell lines than WGD^– cell lines (Fisher’s exact test, p < 0.1) in a specific tumor subtype, it was considered a “hit” in this analysis (Extended Data Fig. 3a).

Within each tumor type, the median essentiality scores for each gene in the WGD^– and WGD⁺ cell lines were identified. When a gene showed a statistically significant enrichment in its median essentiality score in the WGD⁺ compared to the WGD^– cell lines (Wilcoxon test, p < 0.05), and also had an essentiality score of ≤ −0.5 in the WGD⁺ cell lines, it was considered at “hit” in this analysis (Extended Data Fig. 3a).

We employed the thresholded analysis with the Fisher’s exact test and non-thresholded analysis with the Wilcoxon rank-sum in each individual tumor type (n=12) as well as in a combined pan-cancer analysis. These analyses were also performed separately for the CRISPR and RNAi datasets. Only genes that had measurable data in 95% of total cell lines were analyzed. The final PSL score for each gene was the total number of instances a gene was found to be a hit across all analyses (Fig. 2b, Supplementary Table 3,4). As a result, some hits may have come entirely from either the CRISPR or RNAi datasets, such as KIF18A which was only found to be enriched for essentiality in the CRISPR dataset, likely due to insufficient knockdown in the RNAi dataset.