Federal Project Grant Award Summary

Yale University received a $819,674 Project Grant from the National Institute of Allergy and Infectious Diseases (NIAID) under the Allergy and Infectious Diseases Research program (CFDA 93.855) to develop mechanistically informed deep learning computational models for characterizing autoreactive B cells in myasthenia gravis (MG). The award, obligated on May 15, 2026, with a completion deadline of April 30, 2031, will fund the development of interpretable artificial intelligence systems using transformer architectures and protein language models to identify pathogenic autoantibodies and abnormal B cell receptor (BCR) sequences that escape detection in seronegative MG patients. This research addresses a critical diagnostic gap affecting approximately 15% of clinically diagnosed MG patients who lack detectable autoantibodies in current assays, thereby delaying diagnosis, treatment initiation, and clinical trial enrollment.

The research will deliver two primary computational and experimental products: (1) interpretable deep learning models incorporating anomaly detection and uncertainty quantification to identify BCR sequences enriched in autoimmune repertoires, with integrated methods to identify molecular features driving model predictions; and (2) evolution-aware BCR models that integrate clonal phylogenies to detect abnormal somatic hypermutation patterns and atypical B cell maturation trajectories. Comparative analysis across healthy individuals, autoimmune disease patients, and both seropositive and seronegative MG cases will produce convergent biomarker signatures specific to autoimmune conditions, ultimately advancing understanding of hidden pathogenic autoantibodies and improving diagnostic capabilities for seronegative MG patients.