Figure 4

Phenotype explanation of 922 CNVs as GDE or TDBD. (A) Counts of DECIPHER deletions classified as GDE or TDBD. It was found that 41 of the TDBD cases did not demonstrate computational evidence of GDE and are indicated as TDBD only in (B). (B) Proportion of CNVs predicted to correspond to TDBD, GDE or both, compared with randomized data by permutation of phenotype annotations of the DECIPHER patients (green) and permutations of phenotype associations of genes to monogenic diseases (purple). 6.72% of the deletions in the DECIPHER cases were predicted to be TDBD or mixed TDBD/GDE, compared to 5.18% on average for randomized (phenotype-shuffled) deletions (P=0.0008) and 4.88% for randomizations with permuted gene phenotypes (P=0.003). A pure TDBD mechanism was predicted for 4.45% of the DECIPHER cases and a mean of 2.84% of the randomizations with permuted CNV phenotypes (P<0.0001), and in 3.02% with permuted gene phenotypes (P=0.014). `No boundary enhancer adoption’ refers to deletions that do not overlap a boundary element but have a matching enhancer and gene signature in the 400-kb flanking regions. Here no significant enrichment over randomized data was observed, suggesting that the disruption of chromatin architecture contributes to TDBD-related enhancer adoption. CNV, copy-number variation/variant; GDE, gene-dosage effect; kb, kilobase; TDBD, TDB disruption.