Funded Research
We invest in scientists who are harnessing the power of computational biology research to accelerate discoveries that will improve health outcomes for millions of people.
.svg){kind=icon}
Computational Biology
Biswas Family Foundation empowers computational biology research, leveraging advanced data analysis, modeling, and algorithms to transform the way we study, diagnose, and treat disease. By harnessing the power of computation, we aim to accelerate discoveries that will improve health outcomes for millions worldwide.
Transformative Computational Biology Grants
In March 2024, five research teams were selected to receive funding for Biswas Family Foundation’s Transformative Computational Biology Grant Program. As part of a $15 million funding round, multidisciplinary research teams were awarded up to $1 million over two years for research using computational tools across a range of focus areas, including artificial intelligence for genomic medicine, diagnosis of cardiovascular disease, precision oncology therapies, enhancement of clinical datasets, and drug repurposing systems.
Artificial Intelligence for Genomic Medicine: Circuitry, Treatment, Personalization
Principal Investigator: Manolis Kellis, PhD, Professor of Computer Science, Massachusetts Institute of Technology
Co-Investigators: Brad Pentelute, PhD, Professor of Chemistry, Massachusetts Institute of Technology, Marinka Zitnik, PhD, Associate Professor of Biomedical Informatics, Harvard Medical School
This project aims to support the integration of single-cell and spatial sequencing with machine learning approaches to predict target genes for precision therapeutics in cancer, neuroscience, and metabolic disorders. The high-throughput interdisciplinary discovery loop will help advance biomarker selection, de novo drug synthesis, and drug repurposing, which are essential for more effective therapeutics.
A Chatbot Assistant for Genetic Diagnosis and Interpretation of Common and Rare Cardiovascular Diseases
Principal Investigator: Anshul Kundaje, PhD, Associate Professor of Genetics and Computer Science, Stanford University
Co-Investigator: Jesse Engreitz, PhD, Assistant Professor of Genetics, Stanford University
This project aims to accelerate the diagnosis of patients with cardiovascular diseases by developing an artificial intelligence chatbot interfaced with genomic knowledge bases. In the clinical diagnostic workflow, the impact of genomic mutations is unknown in many cardiovascular diseases, so chatbots with access to genomic knowledge bases populated by machine learning models trained on multiple types of data will improve the diagnostic process.
Biswas Center for Transformative Computational Cancer Biology
Principal Investigator: Katherine Pollard, PhD, Director of Data Science & Biotechnology, Gladstone Institutes
Co-Investigators: Alex Marson, MD, PhD, Barbara Engelhardt, PhD, Catherine Tcheandjieu Gueliatcha, DVM, PhD, Christina Theodoris, MD, PhD, Karin Pelka, PhD, Ryan Corces, PhD, Seth Shipman, PhD, and Vijay Ramani, PhD
This project aims to support the development of personalized diagnosis and treatment for colorectal and skin cancers. Machine learning models will be trained to predict how a patient’s genetic mutation can alter tumor cell biology to better understand the diverse pathways that drive cancer, and to evaluate the effectiveness of candidate immunotherapies for more personalized treatments.
The MAIDA Initiative: Democratizing Global Medical Imaging Data Sharing for Safer and Fairer Artificial Intelligence
Principal Investigator: Pranav Rajpurkar, PhD, Assistant Professor of Biomedical Informatics, Harvard Medical School
This project aims to collect medical imaging data from around the globe to facilitate the use of artificial intelligence to analyze images and improve diagnosis and treatment. The focus is on collecting chest X-rays and chest CT images for a variety of clinical settings. Developing an open data repository of medical images that is representative of diverse patient populations will help the deployment of artificial intelligence tools that are reliable, equitable, and inclusive.
CURE-Bench: Universal Benchmark for All-Disease Drug Repurposing
Principal Investigator: Marinka Zitnik, PhD, Associate Professor of Biomedical Informatics, Harvard Medical School
This project aims to build CURE-Bench, a comprehensive all-disease benchmark for evaluating computational drug repurposing systems. An international competition will help establish common tasks and datasets to promote the development, evaluation, and widespread use of artificial intelligence models to identify clinically-relevant drug hits. The goal is to develop foundation artificial intelligence models and an accompanying evaluation framework to benchmark models across diseases.
Biswas Fellows
Launched in 2025 by the MIT Health and Life Sciences Collaborative (MIT HEALS), the Biswas Postdoctoral Fellowship program provides four years of salary support to exceptional early-career postdoctoral researchers at MIT. By supporting cutting-edge research in the health and life sciences, the program aims to transform healthcare and improve the lives of millions worldwide.
Fast Grants
Established in 2026, Biswas Family Foundation Fast Grants are lightweight, fast-turnaround gifts for researchers who want to explore an idea at the intersection of AI and health. This program aims to catalyze high-ambition pilot projects for early-career scientists or established researchers pivoting into ambitious new fields.
Awarded amount: 100k
Basis of a Streaming AI Controller for Evidence-Driven, Low-Cost Cancer Sequencing
An open-source streaming AI that calibrates long-read tumor calls, ranks the actionable ones, and decides when enough has been sequenced, cutting cost and review time in cancer diagnosis.
Discovering the Molecular Trace of Alzheimer's Disease on Routine Brain MRI
Pathologic tau protein drives Alzheimer's disease, but there is no scalable, widely available technique to measure regional tau burden. We are testing whether explainable AI can assess tau from MRI scans already obtained.
GenVinci-Dx: AI-powered 3D Spatial Transcriptomics for Cancer Diagnosis from Routine H&E Sections
GenVinci-Dx reconstructs 3D single-cell spatial transcriptomic cancer profiles from routine H&E slides, enabling spatial molecular diagnostics without sequencing.
AI-driven simplification of highly-multiplexed protein imaging for clinical deployment
We will use AI to convert complex spatial protein imaging into simpler, clinically deployable panels that predict therapeutic response.
Improving epilepsy treatment through AI-enhanced brain images
We used AI to enhance the quality and resolution of brain images, and we will test whether clinicians can use these enhanced images to better plan targets for surgery in a real-life epilepsy clinic.
Passive multimodal forecasting of nocturnal hypoglycemia in type 1 diabetes
AI that fuses CGM and smartwatch signals to predict dangerous overnight low blood sugar in type 1 diabetes during sleep.
Awarded amount: 50k
An Agentic AI-scientist for Prioritizing Dementia Biomarkers
Build an AI-scientist that automatically discovers biomarkers from large-scale dementia proteomics data and reviews literature to prioritize reliable, interpretable follow-up druggable targets.
Cost-aware active diagnostics to prevent glaucoma sight loss
An active-machine-learning policy will use eye scans, visual-field tests, eye pressure, and medical history to catch preventable glaucoma sight loss earlier with sparse, cost-aware monitoring.
An AI co-scientist for spatial-proteomics biomarker discovery
An autonomous AI co-scientist that explores spatial proteomics datasets end-to-end — hypothesises, runs analyses, iterates — to discover candidate biomarkers at scale.
QUANTUM: Quantitative tumour mapping for cancer surgery
QUANTUM will create fluorescence lifetime (FLT) imaging - pathology dataset and AI models to estimate microscopic tumour burden from macroscopic images, enabling better tumour and margin assessment.
Training Splicing Foundation Models for Out-of-Distribution Disease Biology
We will train an open source genomics foundation model that will achieve broad generalization across all RNA splicing data to discover novel biomarkers across neurodegeneration and other diseases.
An AI-Enabled ECG Image Platform for Global Detection of Critical Congenital Heart Disease
To develop and deploy an AI-enabled ECG image platform for detection of critical congenital heart disease (CCHD), enabling scalable neonatal cardiac screening in resource-limited settings worldwide.
A Multiscale Dataset of Alzheimer’s to Unify Understanding from the Molecular Level to Disease
Assemble the first Alzheimer’s dataset linking single-molecule characterisation to tissue-scale pathology in human brains and extract features to enable AI-powered causal inference across scales.
AI-integrated cell-free DNA fragmentomic analyses for predicting immunotherapy response
AI framework turns cfDNA fragment patterns into a minimally invasive blood test that predicts immunotherapy response & demonstrates how integrated AI enables real-time, cost-effective cancer care.
Early Detection of Clinically Significant Prostate Cancer with Multimodal AI
The project aims to build multi-modal AI that combines prostate MRI images, clinical notes, radiology reports, and biopsy results to better predict clinically significant prostate cancer over time.
Awarded amount: 25k
EpiWatch AI: Forecasting Infectious Disease Outbreaks 2–6 Weeks Early in Africa
An open-source AI platform fusing disease surveillance, climate, and mobility data to forecast cholera, Lassa, mpox, and dengue outbreaks 2–6 weeks before standard detection across Africa.
PhageHost-Bench: An Open Benchmark for AI-Driven Phage–Host Prediction
PhageHost-Bench will be the first open, public benchmark and leaderboard for AI prediction of which phage kills which bacterial strain, built entirely from public genomic data.
Mapping the Molecular and Clinical Consequences of X-linked Genetic Variation Across Cancers
Developing an AI-assisted framework to characterize how genetic variation on the X chromosome influences cancer risk and sex-biased cancer outcomes
The Known Unknowns: Accelerating Discovery by Mapping Gaps in the Medical Literature
A platform that completes systematic reviews of medical evidence in minutes, powering a discovery engine for unstudied medical questions
Resolving MYH7 Missense Variants of Uncertain Significance (VUS) in Hypertrophic Cardiomyopathy via Multimodal Graph Networks
We are building AI that reads heart scans and DNA together to determine whether uncertain genetic changes cause hypertrophic cardiomyopathy, a common inherited heart-muscle disease
Fiber-optic AI Guidance for Precision Cardiovascular Device Positioning
Fiber-optic sensing and AI gives heart surgeons an augmented-reality (AR) 3D map of their tools and anatomy for placing implants more precisely, cutting complications and reliance on elite surgeons.
Generative Foundation Models for Arbovirus Emergence Forecasting: A 12-Month Pilot
A 12-month pilot extending ArboFM, an existing arbovirus foundation model, into a generative forecaster that proposes the viral sequences most likely to emerge next.
Autonomous Ultrasound Agent for Imaging and Surgical Navigation
Develop an AI imaging agent that actively acquires and interprets ultrasound data during surgery, enabling more autonomous, scalable, and accessible image-guidance workflows.
Interpretable AI for Cell Microscopy
This project builds AI methods that not only recognize patterns in cell microscopy images but also explain them, connecting high-dimensional image data to biological mechanisms that matter for understanding disease.
AI-Driven Design of a Conformation-Specific Therapeutic for Multiple System Atrophy
Designing conformation-specific protein binders that selectively target and eliminate MSA α-synuclein fibrils to enable a new class of precision neurotherapeutics.
Deep-Learning Model Using cfDNA Fragmentation to Forecast Lineage-Plasticity–Mediated Immunotherapy Resistance in Lung Cancer
We are training a deep-learning model on cfDNA to detect hidden lineage-plastic tumor states that predict poor immunotherapy response.
AI-Native Human Genetics Knowledge Base for Causal Target Discovery
Build a prototype AI-readable knowledge base that turns scattered genetics and proteomics evidence into actionable target-prioritization summaries for human disease.
OpenPath: A Low-Cost Slide Scanner to Deploy AI Cervical Biopsy Triage in Resource-Limited Settings
OpenPath, a <$500 open-source slide scanner (OpenFlexure + 40x) for biopsies, paired with our PAVE AI Cervical Pathology model (PAVE-CP), can increase pathology access in low-resource settings.
An open and accessible AI voice-detector for early dementia
We will develop a deep-learning pipeline that can automatically detect early dementia from a short voice recording, all while protecting data privacy and being deployable on local devices.
Predicting Therapeutic Targets in Aneuploid Cancers Through Virtual Cell Modeling
Using AI-powered virtual cells, we'll predict how chromosome abnormalities promote cancer progression and reveal new treatment opportunities.
A Self-Calibrating Statistical Inference Framework for Trustworthy AI in Single-Cell and Spatial Multiomics
A statistical inference framework that calibrates the outputs of pretrained generative models for single-cell and spatial multiomics, with an agent-based implementation for downstream analysis.
Organizational Support
We give to groups that advance computational biology, foster scientific curiosity, and promote collaboration.
Arc Institute
United States
As part of the $15 million in funding announced in March 2024, the Foundation provided a gift to the Arc Institute to enhance artificial intelligence and computational biology research.
Arc Institute is a nonprofit research organization whose mission is to accelerate scientific progress and understand the root causes of complex diseases. By doing this, they aim to improve human health by narrowing the gap between discoveries and impact on patients.
.avif)
