Clinical genomics for rare pediatric disease
Methods and pipelines for structural and copy-number variant detection in clinical genome sequencing, built with clinical geneticists and genetic counselors to raise diagnostic yield.
Bioinformatics Scientist II · Children's Hospital of Philadelphia · Philadelphia, PA
Tejas R. Jammihal
Bioinformatics scientist building computational and machine learning methods for clinical genomics and cancer immunogenomics.
About
I'm a bioinformatics scientist at the Children's Hospital of Philadelphia, where I develop and clinically translate methods for finding structural and copy-number variants in genome sequencing data to help diagnose rare pediatric disease. Before that, I spent several years in cancer immunogenomics at Dana-Farber Cancer Institute and MD Anderson Cancer Center, studying why some patients respond exceptionally well to immunotherapy. That work led to co-first-author papers in Nature Cancer and Cancer Cell. I started my career in pharma R&D as a computational biology co-op at Novartis.
I like working end to end, from scoping a question with clinicians and scientists to building pipelines and models that change how they work.
Currently working on: structural variant detection and machine learning-based variant triage in clinical genome sequencing.
What I work on
Cancer immunogenomics
Why some patients respond exceptionally well to immune checkpoint inhibitors: clonal neoantigens, tertiary lymphoid structures, and molecular subtypes across phase III trials.
Machine learning for genomic interpretation
Deep learning and AutoML for triaging variant calls, with human-in-the-loop design so that no confirmed finding is dropped.
Selected work
All publications →-
Immunogenomic determinants of exceptional response to immune checkpoint inhibition in renal cell carcinoma
Clonal neoantigen load and tertiary lymphoid structures mark exceptional responders where tumor mutational burden, the standard biomarker, does not.
-
Impact of renal cell carcinoma molecular subtypes on immunotherapy and targeted therapy outcomes
Transcriptomic subtypes transferred between two phase III trials with a random-forest classifier, validating their biology across cohorts.
-
Overcoming a diagnostic blindspot: identifying balanced translocations in Mendelian rare disease cohorts
A high-specificity pipeline that resolves balanced reciprocal translocations from short-read genome sequencing, applied across five cohorts and 16,000+ genomes, with new diagnoses in previously unresolved cases.
-
Multi-omic re-analysis increases diagnostic yield in individuals with Cornelia de Lange syndrome
Integrating genome and transcriptome data solved cases that single-assay analysis had left unresolved.
On the side
I use AI coding tools daily in my research work, and I like building small things with them outside of it.
Training Log
A local-first workout app built around one idea: zero decisions to start. Open it and today's workout is already planned for the time you have and the equipment you own, personalized by an LLM behind a small proxy, with a curated exercise library as fallback. Works offline and signed out, and syncs across devices when you sign in.
Get in touch
Open to collaborations on clinical genomics, structural variation and cancer immunogenomics. Email hello@tejasrj.io.