Exemplar Project: Matchmaker Exchange

Transcription

Exemplar Project: Matchmaker Exchange
Exemplar Project: Matchmaker Exchange Heidi L. Rehm, PhD Partners Healthcare, Brigham & Women’s Hospital and Harvard Medical School hBp://genomicsandhealth.org/our-­‐work/iniHaHves/matchmaker-­‐exchange Genomic Matchmaker Pa#ent #1 Clinical Gene#cist #1 Pa#ent #2 Clinical Gene#cist #2 NoHficaHon of Match Phenotypic Data Feature 1 Feature 2 Feature 3 Feature 4 Feature 5 Genotypic Data Gene A Gene B Gene C Gene D Gene E Gene F Genomic Matchmaker Genotypic Data Gene D Gene G Gene H Phenotypic Data Feature 1 Feature 3 Feature 4 Feature 5 Feature 6 Courtesy of Joel Krier Matchmaker Par#cipants Gene Matcher ClinGen GEM.app DECIPHER Matchmaker Exchange Phenome Central LOVD Café Variome Gene Yenta Undiag. Diseases Program PaHent-­‐to-­‐ PaHent Matching Applica#on Programming Interface (API) Version 1 API •  Based on PhenomeCentral and GeneMatcher API •  Data requester reveals a set of paHent data to a ME database •  Control over the matching algorithm is leZ to the sites •  hBps://docs.google.com/document/d/
1sDp_IYenhk3kdCCEEcFL_GAPVJqpB0TDPNUC9pZ4xo8/edit • 
• 
• 
• 
• 
• 
• 
• 
• 
ID (Mandatory) -­‐ The internal idenHfier (obfuscated or not) used by the originaHng system to reference the paHent data. Label (OpHonal) -­‐ A name/idenHfier assigned by the user which can be used to reference the paHent in a recognizable manner (in an email for example); it should not contain any personally iden:fiable informa:on. Query type (OpHonal) -­‐ Accepted values: once, periodic SubmiBer (Mandatory) -­‐ Consists of contact informaHon of the person that submiBed the search Gender (OpHonal) Age of onset (OpHonal) Mode of inheritance (OpHonal) Disorders (OpHonal) -­‐ A list of OMIM (MIM:######) or OrphaNet (ORPHA#####) idenHfiers Features – It is mandatory to have at least clinical features or gene(s ); having both is preferred – 
– 
HPO terms for clinical features gene name(s) Steps and Policies for Matchmaking •  SubmiBer deposits* case with at least : –  gene candidate(s) and/or –  disease name (OMIN or Orphanet IDs) or high-­‐level HPO terms (more detailed HPO terms with consent) –  submiBer idenHty+ •  Query# of all API-­‐linked databases •  If match occurs, depositor and requestor are both noHfied – details of case are shared manually to begin follow-­‐up studies to validate match *SHll under debate: Requirement for data submission vs query without submission +ME Policy requires submiBer to be idenHfied #Recommended best pracHce -­‐ each ME site keeps an auditable record of all queries Matchmaker Exchange: GA4GH Working Group Interac#ons Data Working Group •  Shared API for review by Data WG •  AcHve discussions on API specificaHons by conference call and GitHub •  hBps://github.com/ga4gh/schemas/pull/137 Regulatory and Ethics Working Group •  Requested guidance on informed consent for different Hers of matchmaking – task force being formed •  Proposal: no consent for gene and high-­‐level phenotype; consent required for sharing variants, genomic files and detailed phenotypes Security Working Group •  Need input on query authenHcaHon and verificaHon steps; security requirements for databases on the exchange Clinical Working Group •  Engaging experHse on phenotyping for matchmaking: Will need improved specificity and scoring for phenotype queries as datasets enlarge PhenomeCentral
›  rapid phenotyping using HPO
›  add VCF and/or top genes
phenomecentral.org
Strongly matched
MFDM patient
despite atypical
presentation
›  557 cases -  deeply phenotyped -  most with exome data
-  most undiagnosed
›  272 users
›  2 validated matches
Two undiagnosed
patients matched,
STIM1 identified as
candidate
(paper in prep)
includes data from:
Courtesy of Orion Buske and Mike Brudno GeneMatcher •  Designed to connect via e-­‐mail clinicians/
researchers with interest in a gene about which nearly nothing is known •  Only a human gene symbol or ID is required •  Can also match on OMIM phenotype numbers •  Matching on phenotypic features is in development As of 10/1/2014 688 genes submiBed by 174 submiBers from 19 countries 15 matches to date No system to track validaHon of matches Courtesy Ada Hamosh, François Schie7eca7e, Nara Sobreira Over 600 publicaHons resulHng from matchmaking in DECIPHER since 2010 Highlights • Flexible and responsible data sharing • HPO Phenotype-­‐linked Sequence and Copy-­‐
Number Variants • 14000+ anonymous records shared publicly/10000+ shared privately • Direct depositor E-­‐Mail and DECIPHER-­‐
mediated matchmaking support • 43 countries/260+ Projects/1400+ registered users • Real-­‐Hme analysis and comparison tools Primary ObjecHve Facilitate iden:fica:on and interpreta:on of phenotype-­‐linked pathogenic gene:c varia:on in rare disorders Courtesy MaF Hurles, Helen Firth and Jawahar Swaminathan Genomes Management Application (GEM.app)!
genomics.med.miami.edu!
•  Free web-based exome/ whole genome pipeline and matchmaking tool!
•  User-controlled sharing of data between individuals, consortia, ad hoc
collaborators!
•  >450 registered users, 80 institutions, 36 countries, >5,000 exomes/1,000
genomes, 90 phenotypes!
•  Results: 53 novel disease genes identified/ 34 published in past 24 months !
Courtesy!
Stephan !
Züchner!
Café Variome: for data discovery+sharing and pa#ent matching in networks and globally Open Discovery – ReporHng Existence of Variants in Sources Open Access Core info for each variant record is shown & made available for download Restricted Access Linked Access Core or full record details are provided per record, if: •  User is pre-­‐approved by group access permissions set by data owner •  Access is approved aZer facilitated email request to the data owner Source DB resource No data, only link to the data source is reported. Access managed by source db -­‐ Allows users to interrogate paHent records based on highly flexible GENO and/or PHENO query opHons (generic API available), to locate ‘exact’ or ‘similar’ hits (‘ontology based’) -­‐ Sits alongside source databases, as a ‘discovery shop window’, allowing complete flexibility over which fields are search or returned, with user/role based permissions managed at single rAnthony ecord level Brookes ClinGen: The Clinical Genome Resource
www.clinicalgenome.org ClinGen is creaHng a case-­‐level repository for: •  Managing bulk submissions to dbGaP on behalf of clinical and research labs •  Controlled access to individual genotype and phenotype data •  Public access to aggregated data (e.g. allele frequencies, beacon queries) •  Matchmaking within repository and API to Matchmaker Exchange •  GenomeConnect paHent registry will support paHent matching Courtesy of Heidi Rehm The UDN is a new ini#a#ve that expands on the NIH Undiagnosed Diseases Program (UDP) The UDP has reviewed 3300 medical records, seen 750 pa:ents with rare and undiagnosed condi:ons, and iden:fied more than 70 rare diseases and several new condi:ons. GOALS OF THE UDN: •  Facilitate diagnosis of currently undiagnosed diseases and condiHons •  Encourage collaboraHon •  Enhance research experience for paHents and families with undiagnosed diseases AS PART OF GENOMIC MATCHMAKER EXCHANGE, UDN PLANS TO: •  Use self-­‐phenotyping surveys •  Collect phenotypic data using HPO •  Issue RFAs and gene funcHon collaboraHon candidate lists •  Educate member sites about Matchmaker Exchange and all that it offers Courtesy of Catherine Brownstein and Ingrid Holm www.LOVD.nl •  Primarily a LSDB
system
•  Can submit variants/
phenotypes for cases
– request VIP status
Nature Genet 46: 188–193 (2014) •  Can unlink V&P to
keep anonymous
(contact through DB curator
if match is made)
Human and Clinical Genetics
© JT den Dunnen
GeneHc Alliance PEER PaHent Registry Plauorm for Engaging Everyone Responsibly (PEER): Enabling families and researchers to FIND DATA, USE DATA, and CONTACT PEOPLE Matchmaker Find other children like my child hBp://geneHcalliance.org/programs/biotrust/peer www.matchmakerexchange.org Launched last night! AcHviHes in Progress •  ConHnue work with GA4GH WGs to finalize API specificaHons, agree on consent policies, and address security requirements •  Work with each ME database on API implementaHon and address any barriers •  Develop guidance for groups without a database wishing to choose a site for data deposiHon and matchmaking support Acknowledgements S Balasubramanian Mike Bamshad Sergio Beltran Agullo Jonathan Berg Kym BoycoB Anthony Brookes Catherine Brownstein Michael Brudno Han Brunner Orion Buske Deanna Church Raymond Dalgliesh Victor de la Torre Andrew Devereau Johan den Dunnen Sergiu Dumitriu Helen Firth Paul Flicek Jan Friedman Richard Gibbs Marta Girdea Robert Green Ada Hamosh Ingrid Holm MaB Hurles Ekta Khurana SebasHan Kohler Joel Krier Owen Lancaster Melissa Landrum Farrah Ladha Paul Lasko Rick LiZon Daniel MacArthur Alex MacKenzie Danielle MeBerville Aleksander Milosavljevic Debbie Nickerson Woong-­‐Yan Park JusHn Paschall Anthony Philippakis Heidi Rehm Peter Robinson Gary Saunders Francois SchieBecaBe Rolf Sijmons Nara Sobreira Jawahar Swaminathan Morris Swertz Peter Taschner Sharon Terry Rachel Thompson Stephan Zuchner Appendix Applica#on Programming Interface (API) The iniHal goal of the Matchmaker Exchange was to establish a Version 1 API and standard operaHng procedure (SOP) to facilitate searching across databases. Version 1 API •  Based on PhenomeCentral and GeneMatcher API •  Involves the data requester revealing a set of paHent data to the matchmaking service provider(s) •  Control over the matching algorithm employed (i.e., the data items considered and the degree of similarity required) is leZ completely in the hands of the matchmaking service •  hBps://docs.google.com/document/d/
1sDp_IYenhk3kdCCEEcFL_GAPVJqpB0TDPNUC9pZ4xo8/edit API v1 Data Fields • 
ID (Mandatory) -­‐ The internal idenHfier (obfuscated or not) that can be used by the originaHng system to reference the paHent data. • 
Label (OpHonal) -­‐ A name/idenHfier assigned by the user which can be used to reference the paHent in a recognizable manner (in an email for example); it should not contain any personally iden:fiable informa:on. • 
Query type (OpHonal) –  Accepted values: •  once: only search once in the current database • 
periodic: repeat the search monthly unHl cancelled, reporHng new and updated matches • 
SubmiBer (Mandatory) -­‐ Consists of contact informaHon of the person that submiBed the search • 
Gender (OpHonal) • 
Age of onset (OpHonal) • 
Mode of inheritance (OpHonal) • 
Disorders (OpHonal) -­‐ A list of OMIM (MIM:######) or OrphaNet (ORPHA#####) idenHfiers, can be empty NOTE: we may want to support other sources later. • 
Features – It is mandatory to have at least clinical features or gene(s ); having both is preferred – 
– 
HPO terms for clinical features gene name(s) Matchmaker Exchange Tiered Informed Consent Proposal •  Level 1 -­‐ Undertaking matchmaking based on top level HPO terms and/or candidate gene names -­‐ consent not required –  Broad phenotype descripHon -­‐ as a disease name (OMIM or Orphanet e.g. Charcot-­‐Marie-­‐Tooth disease) or by using top level HPO terms (axonal motor neuropathy) –  HGNC approved gene names -­‐ for the suspected or candidate pathogenic loci –  This level of informaHon is essenHally non-­‐idenHfiable and therefore its use in data discovery (or even data sharing) implies no significant risk of harm and so such uses should not require explicit paHent consent –  Subsequent exchange of this level of informaHon between depositor and requester (i.e., data sharing) raises no addiHonal consent risks –  Subsequent exchange of more detailed informaHon between depositor and requester (i.e., data sharing) may require consent (see below) •  Level 2 -­‐ Undertaking matchmaking based on any depth of HPO terms and/or DNA or protein sequence level informaHon including genomic variant datasets -­‐ consent required –  Detailed phenotype descripHon -­‐ using any set of HPO terms –  Genomic variant dataset -­‐ including one or more variants (irrespecHve of suspected eHologic role), or related informaHon such as variant class, amino-­‐acid alteraHon, variant locaHon, affected exon, etc. –  This level of informaHon is idenHfiable and therefore its use in data discovery implies a possible risk of harm and so such uses would require explicit paHent consent –  Subsequent exchange of this level of informaHon between depositor and requester (i.e., data sharing) likewise would require consent –  However, consent for matchmaker discovery and subsequent sharing of level 2 data can be taken to have been given if consent exists for the data to be included in an open or controlled access database whose declared purpose involves data sharing for purposes consistent with matchmaking