Amit Majithia Lab
- Mission
- Research
- People
- Publications
- Data & Code
- Contact
Research
Insulin resistance is a major cause of the epidemic diseases of our society: diabetes, heart attacks, strokes, and fatty liver. Our goal is to understand who develops insulin resistance, how, and why. We use human genetics, functional genomics, and cell-based assays to address this goal. In addition to this discovery science, we focus on clinical translation by developing ML/AI-driven clinical tools to interpret large scale genetic, genomic, and longitudinal health data to diagnose and treat people with diseases related to insulin resistance.
Insulin Resistance Gene and Biomarker Discovery | Metabolic Disease Subtyping for Precision Medicine | Developing Clinical Tools to Provide Diabetes Care
Insulin Resistance Gene and Biomarker Discovery
Insulin resistance remains clinically challenging to quantify and treat. The recent accumulation of genome sequences, serum samples, and clinical characterizations of large patient populations presents a unique opportunity to determine the genetic and molecular underpinnings of insulin resistance. We leverage these resources to identify insulin resistance genes using high throughput cell morphological screens (a) and advanced genetic analyses including Mendelian Randomization (b). In parallel, we apply high throughput ‘omics’ on patient serum/plasma samples to rapidly identify and credential novel biomarkers for clinically or genetically defined subgroups of individuals (c).
a. Jiao Y, et al. Discovering metabolic disease gene interactions by correlated effects on cellular morphology. Mol Metab. 2019 Jun; 24:108-119. Epub 2019 Mar 13. DOI. PDF
b. DeForest N, et al. Genome-wide discovery and integrative genomic characterization of insulin resistance loci using serum triglycerides to HDL-cholesterol as a proxy. Nature Communications. 2024 Sep 14;15(1):8068. DOI. PDF
c. Begzati A, et al. Plasma Lipid Metabolites, Clinical Glycemic Predictors, and Incident Type 2 Diabetes. Diabetes Care. 2025 Jan 6;48(3):1-8. DOI. PDF
Metabolic Disease Subtyping for Precision Medicine
Diabetes is a heterogeneous disease with a polygenic basis. It is difficult to predict the severity of disease and patient risk of microvascular complications like kidney failure and macrovascular complications like heart attack and stroke. To differentiate individuals at high risk for serious complications from those with a more stable disease course, we are conducting longitudinal analyses of the Diabetes Prevention Program (DPP) and its outcomes study (DPPOS), a rich source of data on patients with diabetes that spans multiple decades (unpublished). We also collaborate with multiple other research groups across UCSD to model individual risk of metabolic disease using TG/HDL as an indicator of metabolic health (a). As a complement to these genetic analyses, functional characterization and pathogenic classification of missense variants in clinically important genes provides molecular diagnoses, allows estimation of risk, and guides therapy. We conduct massively paralleled variant characterization of genes with known roles in metabolic disease, including PPARG (b) and HNF1A (c).
a. Lee I, et al. Mechanistic genotype-phenotype translation using hierarchical transformers. bioRxiv 2024.10.23.619940. 2024. DOI. PDF
b. Majithia AR, et al. Prospective functional classification of all possible missense variants in PPARG. Nature Genetics. 2016. DOI. PDF
c. DeForest N, et al. Human gain-of-function variants in HNF1A confer protection from diabetes but independently increase hepatic secretion of atherogenic lipoproteins. Cell Genomics. 2023 3(7) 100339. DOI. PDF
Developing Clinical Tools to Provide Diabetes Care
Providing comprehensive care to people with diabetes both at home and in the hospital requires significant clinician and patient resources. For example, continuous glucose monitors (CGM) used at home provide a high resolution look at an individual's glycemic patterns, but generate over 250 datapoints per day and require patient education and support to ensure optimal self-management of blood sugar. Towards this barrier, we have collaborated within a team of clinicians and researchers to develop the Onduo virtual diabetes clinic to provide virtual clinician support, lifestyle coaching, and a CGM device to patients with diabetes. Implementation of the virtual clinic significantly improved patient satisfaction and reduced HbA1C levels (a), and also led to optimization of diabetes medication management (b). Specifically during the COVID-19 pandemic, we also evaluated the use of CGM for critically ill inpatients with diabetes and COVID-19 and were able to reduce clinician-patient contact and effectively manage glycemic control (c).
In the hospital, the gold standard for glycemic management requires daily individualized insulin dosing decisions. While each decision is usually made according to standard algorithms, hundreds of such decisions must be made daily by a few providers, leading to burnout. We are developing a clinical decision-making tool to address this burden. aiDose, a foundation model for inpatient diabetes management, will incorporate medical records and previous clinician notes to provide decision support for inpatient diabetes services.