project-idea
POU2F1 acts as the master transcriptional activator of DLL3 in SCLC — meaning DLL3 expression levels, which determine whether patients respond to DLL3-targeted therapies like tarlatamab, are now mechanistically explainable and potentially predictable from upstream transcription factor activity.
project-idea
FDA-approved SSRIs synergize with CDK9 inhibitors to kill diffuse midline glioma cells — meaning drugs that are safe, cheap, brain-penetrant, and already FDA-approved may have immediate translational potential in a childhood brain cancer with no effective treatments.
project-idea
This study created the first humanized ossicle model that faithfully reproduces myelofibrosis features including reticulin fibrosis and osteosclerosis, and identified SPP1 as a new therapeutic target — highlighting both a model gap and a drug target gap that a structured resource could help address.
project-idea
BRCA1/2 mutations in lung adenocarcinoma predict better responses to immune checkpoint blockade despite worse overall prognosis — creating a paradox that clinicians need tools to navigate when making immunotherapy decisions for NSCLC patients.
project-idea
Transformed breast cells use physical nanotube connections to transmit BMP signaling to normal neighbors, actively spreading a preneoplastic state — meaning early cancer initiation may be a contagious process, not just isolated mutations.
project-idea
When PI3K is blocked in AML cells, the cancer compensates by upregulating EZH1 — an escape route that can be closed by combining PI3K inhibitors with EZH1/2 dual inhibitors. This research makes a compelling case for systematically mapping known resistance bypass routes and their combination therapy…
project-idea
AR-Status Dashboard: Androgen Receptor Pathway Deactivation Tracker for Prostate Cancer Loss of canonical AR signaling correlates with prostate cancer progression, raising the concern that androgen deprivation therapy may paradoxically promote escape in tumors where AR is already deactivated. Build a clinical evidence aggregator and visualization dashboard that synthesizes published data
project-idea
ProteostasisVuln: A Cancer Proteostasis Vulnerability Screener Cancer cells with low HSF1 expression become catastrophically sensitive to mTORC1 reactivation, creating a potential synthetic lethality. This selective vulnerability does not exist in normal cells, suggesting it could be mapped across many cancer types. Build an open bioinformatics tool that cross-references public tumor
project-idea
ChemoSense TNBC: A Pre-Treatment Biomarker Screener for Triple-Negative Breast Cancer Chemoresistance Pre-treatment biopsies from TNBC patients show distinct proteomic and metabolomic signatures — particularly SIRT5 overexpression — that predict de novo chemoresistance before a single dose of treatment is given. The ability to predict chemotherapy resistance before treatment starts is one of
project-idea
PedBrainTarget: A Drug-Target Interaction Tracker for Pediatric Brain Cancer Stem Cell Pathways A REST-AKT-SOX2 regulatory axis maintains therapy-resistant stem/progenitor cells in Sonic Hedgehog medulloblastoma — and both AKT inhibitors and Midkine inhibitors with clinical-stage compounds already exist. Pediatric brain tumors are difficult to treat partly because of the disconnect between
project-idea
MOC Drug Compass: A Patient-Facing Drug Evidence Explorer for Mucinous Ovarian Carcinoma The first comprehensive drug efficacy dataset for mucinous ovarian carcinoma was just generated using patient-derived tumor organoids — for a cancer subtype that has almost no clinical evidence base and is routinely treated with regimens designed for a different
project-idea
TLS-Scope: An Open-Source Spatial Transcriptomics Pipeline for Tertiary Lymphoid Structure Classification A hierarchical graph neural network operating on spatial transcriptomics data can classify tertiary lymphoid structures as immunogenic or tolerogenic — a distinction with direct implications for predicting immunotherapy response that bulk RNA analysis cannot make. The AI model described in