DECIPHER: Supporting the interpretation and sharing of rare disease phenotype-linked variant data to advance diagnosis and research

Abstract

DECIPHER (https://www.deciphergenomics.org) is a free web platform for sharing anonymized phenotype-linked variant data from rare disease patients. Its dynamic interpretation interfaces contextualize genomic and phenotypic data to enable more informed variant interpretation, incorporating international standards for variant classification. DECIPHER supports almost all types of germline and mosaic variation in the nuclear and mitochondrial genome: sequence variants, short tandem repeats, copy-number variants, and large structural variants. Patient phenotypes are deposited using Human Phenotype Ontology (HPO) terms, supplemented by quantitative data, which is aggregated to derive gene-specific phenotypic summaries. It hosts data from >250 projects from ~40 countries, openly sharing >40,000 patient records containing >51,000 variants and >172,000 phenotype terms. The rich phenotype-linked variant data in DECIPHER drives rare disease research and diagnosis by enabling patient matching within DECIPHER and with other resources, and has been cited in >2,600 publications. In this study, we describe the types of data deposited to DECIPHER, the variant interpretation tools, and patient matching interfaces which make DECIPHER an invaluable rare disease resource.

Keywords: Matchmaker Exchange; genetic disorders; genomic medicine; genotype phenotype correlation; rare diseases; variant interpretation; whole-exome sequencing; whole-genome sequencing.

Figure 1

(a) The DECIPHER community is a global network of academic clinical centers with expertise in genetics. Depositing centers are able to send messages directly to other registered users about patient matches through DECIPHER. Since October 2014 over 4500 such messages have been sent. Here, each line represents a collaboration request sent between depositing centers. Unregistered users’ messages, sent through DECIPHER, are not included in this image. (B) The DECIPHER database currently openly shares approx. 40,000 rare disease patient records, built up over time

Figure 2

DECIPHER supports the deposition and sharing of almost all types of genetic variation

Figure 3

All genomic data is visualized in GRCh38, but deposition is still supported in GRCh37/hg19. Tools are provided to visualize the differences between assemblies. These include comparative genome browsers and gene lists for variants lifted over by DECIPHER, and a liftover mapping genome browser track

Figure 4

(A) DECIPHER enables the deposition of phenotypes using HPO terms. (B) DECIPHER supports the deposition of developmental milestones and anthropometric measurements, for example, occipitofrontal (head) circumference

Figure 5

(A) DECIPHER has developed a protein browser that summarizes genotypic data. Tracks include: Pfam domains, DECIPHER and ClinVar variants, gnomAD variants, and region of predicted nonsense‐mediated decay (NMD) escape. (B) DECIPHER supports the annotation and sharing of sequence variant pathogenicity assessments using ACMG guidelines. A pathogenicity evidence interface is available for depositors. Relevant criteria are selected by clicking on the criteria displayed on the left under “Available evidence types.” “Selected criteria” are displayed on the right, along with “Evidence to consider.” “Further information” links provide recommendations for the use of criteria. In this example, a variant in SLC9A6 is being annotated and ClinGen Variant Curation Expert Panel specifications exist for this gene. Detailed information about these recommendations are displayed by clicking on the “Gene recommendation” links—expert panel recommendations for de novo criterion PS2 are displayed. As criteria are added, DECIPHER calculates the variant pathogenicity according to criteria‐combining rules detailed in the original 2015 guidelines, and according to the ClinGen SVI Working Group's Bayesian classification framework. (C) DECIPHER supports the annotation of copy‐number variants according to ACMG/ClinGen technical standards. Similar to the sequence variant interface, “Available evidence types” are displayed on the left, with “Selected evidence” and “Evidence to consider” displayed on the right. As criteria are selected, the classification score and pathogenicity are calculated and displayed at the bottom of the interface. (D) An assessment interface is provided which is designed to be used in a multidisciplinary team meeting to evaluate whether one or more variants explain the clinical features seen in a patient, and record if a diagnosis has been made (or excluded). Depositors can report several lines of evidence, to weigh evidence for or against a genotype‐phenotype relationship. An OMIM gene‐disease pair and assertion is recorded

Figure 6

(A) Quantitative phenotype data (such as developmental milestones or anthropometric measurements) is recorded in DECIPHER and aggregated on a gene‐by‐gene basis. The data is shared openly in a series of graphs that displays expectations for the healthy population (Normal), the DECIPHER population as a whole, and the gene‐specific data. For certain genes, such as EP300 (displayed here), there are composite faces, which highlight facial dysmorphologies. (B) The matching patient interface allows users to view DECIPHER records that overlap a deposited copy‐number, sequence, or insertion variant, or a gene. In this example, the matching patients overlap EP300. Summary information is shown in a series of pi charts, along with phenotypes present in multiple matching patients. The individual patient records are displayed at the bottom of the interface. Filters are available to assist in finding the most relevant patient matches. (C) Within DECIPHER, aggregated phenotype data is used to identify the most discriminating phenotypes associated with disease genes. A table shows the percentage of phenotyped patients with sequence variants in a gene of interest, with a particular phenotype, compared with the percentage of phenotyped patients in DECIPHER with the same phenotype. The odds ratio and p value from Fisher's exact test are displayed. In this example, data for KMT2A is displayed and sorted by p value. (D): Users with write access to an open‐access patient record are able to query the MatchMaker Exchange to search for potential patient matches. DECIPHER is currently connected to Broad‐seqr, GeneMatcher, MyGene2, PhenomeCentral, and RD‐Connect. Details of potential patient matches are displayed within DECIPHER (patient IDs have been removed in this example)

Figure 7

(A) Since its inception in 2004, DECIPHER has been cited in more than 2600 publications. (B) The genes with the most open‐access sequence variants in DECIPHER (at the time of writing). (C) DECIPHER openly shares variants of unknown significance identified in undiagnosed probands in the Deciphering Developmental Disorders study (research variants). For each variant, a page provides details of the variant and high‐level phenotype terms. The number of patients with each variant in the DDD data set is displayed, in addition to the number of patients identified in the GeneDx and Radboud University Medical Center de novo variant data set as described by Kaplanis et al. (2020)