Improvements to the LinkML framework to support the Phenomics First open science resource

点击翻译按钮获取中文摘要

A Phenomics-First Resource (PFR) for interpretation of variants Genomics is key to precision medicine; however, despite the ease of sequencing, clinical interpretation is still thwarted because relevant data (disease, phenotype, and variant) is complex, heterogeneous, and disaggregated across sources. Moreover, this evidence is sometimes incomplete, conflicting, and erroneous. Consequently, clinicians face long lists of candidate diseases, genes, and countless variants of unknown significance. This situation will not improve without capturing and harmonizing the underlying phenotypic information; computability of this information is the bedrock for the emerging field of ​phenomics​. From basic science to clinical care, communities need structured ways to represent and exchange phenotypes and disease definitions. Addressing these fundamental phenomics needs makes it possible to computationally assess and reveal links between diseases and variants. We have previously shown how the addition of phenotypic information using the Human Phenotype Ontology (HPO) can improve the diagnostic yield for hard-to-diagnose patients, and HPO is therefore now a global standard for “deep phenotyping”. We have demonstrated the applicability of deep phenotyping in the evaluation of rare diseases which have overlapping mechanistic underpinnings with common/complex diseases as well as evolutionarily conserved mechanisms in model organisms. Having coordinated the community and prototyped the underlying computational platforms, we will now align both phenotype ontologies and clinical terminologies, enabling better comparison and inference of phenotypes for improved diagnostic efficacy. We propose to develop a Phenomics-First Resource (PFR). ​Specifically we will: 1. Create a community-driven framework of interoperable phenotype definitions across species​ (uPheno) 2. Harmonize human disease definitions with the ​MONDO​ disease alignment resource 3. Create a community-wide exchange standard for clinical and model-organism phenotypes (​Phenopackets​) 4. Develop an integrated phenomics platform ​to provide the research ​(e.g. BioLink) and clinical (​FHIR​) communities with programmatic access to phenomics ontologies, data, and algorithms The dynamic suite of interlinked technologies will together leverage community-developed knowledge in order to make variant interpretation more reliable, better provenanced, and more clinically actionable.