Of these 12,442 SV events, 1847 involved lower expression of the gene by both protein and mRNA (<−0.4 SD from the sample median for each analyte), representing a highly significant overlap between loss-of-function SV events and low gene expression events (3E-10, chi-squared test, Fig. 2a and Supplementary Data 3 ) and illustrating the importance of incorporating expression data loss of function predictions.
3
—
—
The sentences
Alternatively, using a more stringent statistical cutoff for protein, at p < 0.01, with a less stringent cutoff for mRNA, at p < 0.05 for the filtered dataset, a relatively higher percentage—33%—of the 446 statistically significant proteins were also significant by mRNA, and so significant protein-level associations were more likely to be significant by mRNA versus checking mRNA-level associations at the protein level, which may reflect in part some of the mRNAs involved not being translated into protein.
When comparing gene-level SV associations common between mRNA and protein, we utilized relaxed statistical criteria to lower false negatives, consistent with previous studies 46 , 70 , e.g., so as not to exclude significant mRNA associations where the association was at least nominally significant for protein.