Medical Genomics Unit Analytic Projects

National Human Genome Research Institute

Investigators

Linked publications, trials & patents

Abstract

Part of this year focused on the ongoing establishment of the Medical Genomics Unit (MGU), though we have been able to start to produce novel research results and further modify our methods to be applied to increasingly diverse datasets. All of these activities involve the analysis of data (primarily clinical data) pertinent to a variety of genetic disorders. First, we successfully recruited several key individuals to be part of the MGU. These included: a staff scientist with specific expertise in deep learning and related analytic and statistical techniques as applied to clinical/phenotypic as well as genomic data; a genetic counselor with experience in collaborative research and clinical practice in both laboratory and patient-facing areas. These individuals joined in late 2020. Several other individuals joined later, when other NIH labs closed, and we have started to accept our first trainees (eg, 1 summer student and 1 post-bacc IRTA). Second, we have initiated a number of related independent and collaborative projects that focus on the use of deep learning and other deep and machine-learning-related-based techniques to analyze different types of datasets (eg, related to skin findings, facial features, electronic health records, and ophthalmologic testing as relates to genetic conditions). These are primarily computational projects, and start with the collection and collation of data, with an emphasis on data cleaning. A major goal of these projects is to build relatively flexible algorithms that can be applied (with some modifications) to a variety of data sources. For example, an image classification algorithm applied to skin disorder findings, can be modified and applied to OCT images related to eye findings in other genetic disorders. These have yielded several early manuscripts, which are either close to publication or in draft form.