On average per comparison of two diploid genomes by CGH, we found that 445 out of 1,098 (40.5%) of the validated CNVs overlapped with 622 out of 20,174 (3.1%) RefSeq genes (including intronic CNVs), altering the structure of 835 out of 30,917 (2.7%) gene transcripts, and directly altering the coding sequence of 323 out of 27,761 (1.2%) messenger RNAs (Table 1). When all samples were considered together, we found that 3,340 (38.8%) of the validated CNVs overlapped 2,698 (13.4%) RefSeq genes (including intronic CNVs), altering the structure of 3,863 (12.5%) gene transcripts, and directly altering the coding sequence of 1,519 (5.5%) mRNAs (Table 1). Over half of the partial gene deletions that encompass exons are predicted to induce frameshifts, and combining these alleles with whole gene deletions identifies unambiguous loss of function alleles for 267 genes (Supplementary Table 1.2).
