The use of Machine Learning to Interpret BCRs and TCRs
Autoimmune diseases are notoriously difficult to diagnose due to clinical and molecular heterogeneity across patients and frequent overlap with other diseases. Methods for diagnosis often require multiple panels of expensive and time-consuming tests. In a proof-of-concept study published in Science, researchers from multiple institutions, including Drs. Judith James and Joel Guthridge, collaborated with Dr. Scott Boyd at Stanford University to develop an innovative framework for diagnosing immunological conditions. The novel method named Mal-ID (machine learning for immunological diagnosis), integrates B and T cell receptor sequencing with protein language models to successfully distinguish between healthy individuals and those with autoimmune diseases (such as lupus and type-1 diabetes), viral infections (including HIV and COVID-19), and recent flu immunization, achieving an area under the curve of 0.98 and significantly outperforming existing methods. This study highlights the potential of BCR and TCR sequencing as a multi-disease diagnostic tool, which, with further validation, could be adapted for clinical use, potentially facilitating early diagnosis and improving patient outcomes. Read the entire article here.


Read more 

Recent findings from OMRF have revealed molecular clues that will allow doctors to separate Sjögren’s syndrome patients into three distinct categories, allowing for more targeted treatment approaches. Read more about this exciting discovery
Read more
Learn more about this research project 
