Genome analysis and knowledge-driven variant interpretation with TGex

Abstract

Background: The clinical genetics revolution ushers in great opportunities, accompanied by significant challenges. The fundamental mission in clinical genetics is to analyze genomes, and to identify the most relevant genetic variations underlying a patient's phenotypes and symptoms. The adoption of Whole Genome Sequencing requires novel capacities for interpretation of non-coding variants.

Results: We present TGex, the Translational Genomics expert, a novel genome variation analysis and interpretation platform, with remarkable exome analysis capacities and a pioneering approach of non-coding variants interpretation. TGex's main strength is combining state-of-the-art variant filtering with knowledge-driven analysis made possible by VarElect, our highly effective gene-phenotype interpretation tool. VarElect leverages the widely used GeneCards knowledgebase, which integrates information from > 150 automatically-mined data sources. Access to such a comprehensive data compendium also facilitates TGex's broad variant annotation, supporting evidence exploration, and decision making. TGex has an interactive, user-friendly, and easy adaptive interface, ACMG compliance, and an automated reporting system. Beyond comprehensive whole exome sequence capabilities, TGex encompasses innovative non-coding variants interpretation, towards the goal of maximal exploitation of whole genome sequence analyses in the clinical genetics practice. This is enabled by GeneCards' recently developed GeneHancer, a novel integrative and fully annotated database of human enhancers and promoters. Examining use-cases from a variety of TGex users world-wide, we demonstrate its high diagnostic yields (42% for single exome and 50% for trios in 1500 rare genetic disease cases) and critical actionable genetic findings. The platform's support for integration with EHR and LIMS through dedicated APIs facilitates automated retrieval of patient data for TGex's customizable reporting engine, establishing a rapid and cost-effective workflow for an entire range of clinical genetic testing, including rare disorders, cancer predisposition, tumor biopsies and health screening.

Conclusions: TGex is an innovative tool for the annotation, analysis and prioritization of coding and non-coding genomic variants. It provides access to an extensive knowledgebase of genomic annotations, with intuitive and flexible configuration options, allows quick adaptation, and addresses various workflow requirements. It thus simplifies and accelerates variant interpretation in clinical genetics workflows, with remarkable diagnostic yield, as exemplified in the described use cases. TGex is available at http://tgex.genecards.org/.

Keywords: Biomedical knowledgebase; Clinical variant interpretation and classification; Exome sequencing; Hamartomatous polyposis; Next generation sequencing analysis; Non-coding variants; Rare genetic diseases; Whole genome sequencing.

Fig. 1

Architecture of variant interpretation in TGex

Fig. 2

The TGex analysis screen (SNVs). The example shown here (and in Fig. 3) is a case of non-syndromic congenital diarrhea [54]. Following WES, the variant with the strongest phenotype implication for “diarrhea” was within TTC37 (L761P), a known gene for trichohepatoenteric syndrome. The discovery of this novel homozygous damaging missense variant was significant for providing an effective diagnosis for a misdiagnosed case. a The main analysis screen is designed to optimally provide the analyst with information and user-interface options. The main analysis area is divided into dedicated tabs for each genetic model used for the analysis, and an additional tab for incidental findings. Each tab is an interactive table where each row represents a variant, and each column depicts a particular variant attribute. The attributes are divided into 7 categories, each category is collapsed by default, showing a subset of critical attributes, with an option to expand. Each column has two interactive functionalities – sorting (by a click on the header) and filtering (clicking on the filter icon to the right). b The Filters and Tools pane summarizes all applied filters for a specific tab in a given analysis. Via this pane, or alternatively via each of the attribute columns in the main analysis screen, filters can be easily added, edited or removed while reviewing the variants. All applied filters are also documented in the Methods section of the final report. In addition to the column filters described above, the pane includes advanced filter options, including predefined gene panels, manually entered gene list filters, VarElect terms used for phenotype prioritization, and Disease frequency used for the allele frequency filter

Fig. 3

Variant analysis and interpretation. This figure shows several views in TGex providing detailed information and useful links to source data, with the goal to improve and hasten expert variant interpretation. a VarElect MiniCards. The extensive gene-phenotype hit-context evidence from the GeneCards knowledgebase is portrayed in the MiniCards. This figure shows selected parts of the MiniCard for the gene TTC37 and the phenotypes used in the congenital diarrhea case. A list of matched phenotypes is shown in red in the top part, followed by extensive gene-centric evidence for queried phenotype association from various GeneCards sections. This is combined with MalaCards-based evidence, similarly showing queried phenotype associations in diseases associated with the gene TTC37, from various MalaCards sections. Search terms are highlighted throughout the text, and links to specific GeneCards/MalaCards webcard positions enable further scrutiny via more detailed evidence exploration within the knowledgebase. b Variant and evidence selection. Several types of marks can be defined per candidate variant by the analyst, upon clicking the ‘Annotate variant’ button located to the left of each variant row. This includes relevance (High, Med or Low), the pathogenicity of the variant, and a free text note. Below, information pieces regarding the variant/gene pathogenicity can be selected, based on VarElect MiniCards and OMIM disease records. The selected variants and their annotations are propagated to the report. c Gene view. A gene-centric summary for the gene TTC37, including associated diseases, mode of inheritance, and pathogenic variants summary, based on OMIM and ClinVar records. d ClinVar information – ClinVar records matching a given variant, including the condition and clinical significance. e ACMG score – Clinical significance based on the ACMG score. Clicking upon the variant clinical significance value shows a detailed view of the data used for the classification.

Fig. 4

The TGex analysis screen (SVs). SV analysis is exemplified by a list of recurrently mutated regulatory elements discovered in a cohort of patients with neurodevelopmental disorders [63]. The highlighted element overlaps the GH17J002188 GeneHancer, an intronic enhancer of the gene SMG6. Remarkably, this enhancer also targets the WDR81 gene (over ~ 476 kb), with a higher VarElect score for the relevant phenotype (neurodevelopmental, “developmental delay”, etc') than SMG6. a The main analysis area for SVs is divided into 3 sections, including the main section listing the SV events (left), the detailed event section (top, right) presenting a detailed view of the list of genes and GeneHancer regulatory elements that are affected by the event, and the genomic view section (bottom, right) allowing visual examination of the genomic context of each event. b Expanded view of the event genomic context. c Clicking on the Phenotype score for a given GeneHancer opens the VarElect MiniCard for the element-gene-phenotype association. At the top part of the MiniCard, evidence describing the GeneHancer and its association with the gene target is detailed. This includes a list of sources for the identification of the element; a list of transcription factors found to have binding sites within the element; a detailed view of the evidence for element-gene associations. Below the GeneHancer details appear the classic gene-phenotype MiniCards as described in Fig. 3. Importantly, the score used for prioritization in the SV module is calculated by combining the GeneHancer confidence score of the element and of the element-gene association, with the classic VarElect gene-phenotype score of the element target gene