Search Stories

1. As a user, with a candidate gene or two, I would like to identify the patients with "harmful" mutations in this gene, or with this gene as a candidate.

   This could be as simple as responding with a table of variants and the patients they are in, with certain filters applied (exomiser harmfulness score, exomiser relevance score, allele frequency in 1000GP, EVS, and/or ExAC).

2. As a user, with a handful of patients with exomes and some phenotypes, I would like to enter these patients in PhenomeCentral and identify matches.

3. As a user, with a particular variant of interest, I would like to find the allele frequency and genotype frequency of a particular variant (e.g. chr1:136138002A>T) across all the patients in the database (e.g. A: 75%, T: 25%; AA: 40%, AT: 40%, TT: 20%).

   Download ExAC data to see examples. This is also exactly what is necessary to be a GA4GH beacon.

4. As a user, I would like to see the top variants for a single person in PhenomeCentral

    For each patient with exome data, 
        display a list of the top N variants based on exomiser score, 
        with possible filters applied (allele frequency in some? database, het/hom, ...?)
   

5. As a user, I would like to see a list of patients that are the most phenotypically similar to this one.

   Rough outline of PhenomeCentral gene-matching algorithm, as it currently stands:

    For each patient A, for each of the most phenotypicially similar patients B:
        We look at the genes G that pass Exomiser filters for both patients 
        (intersection of the gene sets for each patient)
        For each gene g in G:
            For every other patient O where g passes the filters:
                If g is mutated in O as much as in A or B, 
                    we penalize the gene score according to the phenotypic similarity 
                    between O and either A or B.