Introduction:
Introduction: Almost 50% of variants in ClinVar are variants of uncertain significance (VUS). Experimentally based functional data can help resolve VUS but it can be difficult and time-consuming to determine if functional data are available for a particular variant. To support this need, functional data have been accepted in ClinVar submissions for several years. However ClinVar’s original data model represented functional data in a way that was inconsistent with current guidelines for variant classification, and it did not include sufficient details about the method or the result. A prototype was developed to improve the representation of functional data in ClinVar submissions, XML files, and web pages.

Methods:
Methods: Monthly meetings were held with the team supporting MaveDB to compare data models and work towards community standards for functional data. Interviews were conducted with ClinVar users, including individuals who have submitted functional data and individuals who use functional data in variant classification. Feedback from these groups was used to develop a prototype for changes to the ClinVar submission template, XML files, and variant (VCV) webpages. The prototype was published on GitHub (github.com/ncbi/clinvar/) to demonstrate these changes to users and solicit feedback.

Results:
Results: A new submission spreadsheet template was developed specifically for submission of functional data without a classification for disease, such as results from MAVEs (multiplexed assays of variant effects). The general submission template was updated to improve how functional data are submitted as part of the evidence for a germline variant classification. For example, a laboratory may submit the results from RNAseq analysis or an enzyme assay that was performed to inform a variant classification. Several new fields were added to improve how functional data are structured in the database. For example, the field “functional effect” is a high-level field to indicate whether the variant is functionally normal, abnormal, or uncertain based on the experimental assay. The existing field “functional consequence” was retained as a more specific description of the consequence of the variant on the transcript or protein function. A second new field is “Assay type”, which was added as a general description for the assay, such as “RNAseq” or “enzymatic activity assay”. The existing field “Method” was retained so that the submitter can provide a more detailed description of their specific assay. A prototype was also developed for assertions about the strength of functional evidence, within the context of a named specification for variant classification.

Conclusion:
Conclusions: The proposed updates will make it easier for ClinVar users to find relevant functional data and understand how to use it in variant classification. The addition of new structured fields for functional data will allow more types of queries in ClinVar. For example, the high-level field “Functional effect” will allow a user to find a set of variants within a gene of interest that are either functionally normal or functionally abnormal for use as a control set of data. A user will also be able to search for all variants in ClinVar that have a particular type of functional data, such as patch clamp analysis or a secretion assay. Additionally, assertions for the strength of the evidence will be piloted for submission by expert panels, so that ClinVar users receive guidance on how to use the results from certain experiments as evidence within specific classification guidelines. As more functional data are generated by both clinical and research laboratories, it’s critical to make the data findable and accessible to variant scientists and clinicians. Presenting functional data in ClinVar makes it easily accessible both to individual users on the ClinVar website and to software systems that process the entire ClinVar dataset or access it programmatically.