Per-gene and per-variant inheritance information?

[buske]

PhenomeCentral collects patient-level inheritance mode as HPO terms, and variant-level inheritance as one of the following strings:

• de novo germline
• de novo somatic mosaicism
• maternal
• paternal
• unknown

I think we should collect this information at the gene level as well, having just observed someone enter "de novo" into the comment box for the candidate gene. It seems important for matching.

The questions are:

• Are any other groups collecting this information?
• Are any other groups interested in matching on this information over the MME?
• If so, what ontology should we use for this information?

The API currently has a patient-level inheritanceMode field, using descendants of HP's mode of inheritance (www.human-phenotype-ontology.org/hpoweb/showterm?id=HP:0000005). This includes AD, AR, Sporadic, etc.

This is used differently from the way people seem to describe the gene-level or variant-level inheritance, which are more like the descendants of SO's variant_origin (http://sequenceontology.org/browser/current_svn/term/SO:0001762). This includes de novo, germline, somatic, maternal, paternal.

[jxchong]

We are collect gene-level inheritance pattern using the same terms we use for variant-level.

de novo
compound het
AD
homozygous recessive
X-linked dominant
X-linked recessive
mitochondrial
Y-linked
Unknown/Other

Neither the sequence ontology nor the HPO inheritance model choices fit quite right vs how data is actually analyzed, although the HPO choices are closer.

[Relequestual]

DECIPHER does per variant level. Have I previously provided this list? I don't have immediate access to it right now.