Compute CDS involvement for variants partially overlapping genes

[tnavatar]

Many CNVs have partial overlap with clinically relevant genes; the kind of overlap will determine whether the CNV is predicted to adversely affect the involved gene:

The search criteria for confident null deletions are:

• More than 50% of CDS deleted
• Deletion of the start codon
• Deletion of the first exon
• Deletion of an odd number of splice signals
• Frameshift

Some of these are going to be harder to judge than others; there is also going to be the question of which transcript to use. In the end we probably need to evaluate this across all known transcripts and report on each separately.

For each exon/segment of coding sequence, we find ourselves asking the same questions we do about genes:

Exon involvement:

• Is the exon completely enveloped by the CNV? Partially? Does the exon completely contain the CNV?

CDS involvement:

• What percentage of the coding sequence is affected by the CNV

Endianness:

• Which end of the gene has partial overlap?

Might be for further work, but do we care about possible upstream regulatory elements?