Matchmaking through the Matchmaker Exchange
DECIPHER is a founding member of the Matchmaker Exchange (MME, https://matchmakerexchange.org), a Global Alliance for Genomics and Health (GA4GH) driver project which enables the federated discovery of similar rare disease patient data in connected databases. This worldwide collaboration allows automated matchmaking of genetic and/or phenotypic data between databases, via an application programming interface (API). Through MME, DECIPHER is currently connected to Broad-seqr (https://seqr.broadinstitute.org/matchmaker/matchbox, Arachchi et al., 2018), GeneMatcher (https://genematcher.org, Sobreiraet al ., 2015), MyGene2 (https://www.mygene2.org/, MyGene2, 2016), PhenomeCentral (https://phenomecentral.org, Buskeet al ., 2015) and RD-Connect (https://platform.rd-connect.eu, Lochmüller et al ., 2018). Since 2020, DECIPHER depositors have made approximately 1,500 requests for matches from connected databases and received details of more than 4,100 potential patient matches. In the same time period, DECIPHER has received more than 55,000 requests for matches from connected databases, and has returned details of more than 255,000 potential patient matches.
Within DECIPHER, users with write access to a patient are able to query the MME. It is essential that the patient record in DECIPHER has explicit consent for open sharing, as some connected databases have dual notification i.e. they provide their user with details of any potential patient match, and unshared patient records will not be available to users of the other databases. Once MME is queried and the connected databases have responded, details of potential patient matches are displayed within the DECIPHER interface. Potential matches from each database are displayed in a tabular format with matching phenotypes in bold, assisting users in determining the level of phenotype similarity (Fig. 6D). DECIPHER supports the querying of MME for patients with at least one open-access sequence or a copy-number variant which overlaps one gene. Other types of variants present in the patient record will not be included in the MME request.
When a MME request is sent to DECIPHER which contains genomic information, all open-access patient sequence or copy-number variants which overlap a single gene, and all DDD consortium research variants (see Driving rare disease research section) are evaluated for similarity based on functional overlap. Many of the variant requests received from connected databases provide genomic coordinates in GRCh37, and in these cases DECIPHER performs liftover to convert the coordinates to GRCh38 prior to identifying matches. A score for each potential patient match is provided, ranging from 0 to 1, with 1 indicating a better match. DECIPHER’s scoring algorithm for genomic matches takes into account the Ensembl VEP predicted consequence, assessing the severity and similarity of the consequence to those provided in the request.
If only phenotypic data is provided, all open-access patients with phenotypes are evaluated for a match. This takes into account all HPO ancestor terms for both the patient in the request, and patients within DECIPHER. These matches are scored by generating an Intersection over Union score comparing the HPO ancestor terms of the request patient and the patient in DECIPHER.
DECIPHER returns the 20 highest scoring matches per MME request. In the case where there are many matches, the patients’ chromosomal sex is taken into account in addition to the score, in order to prioritise the best possible matches. The returned matches include variant, phenotype (including absent phenotypes) and diagnosis information.