Challenges and Opportunities in Intelligent Molecular Medicine
summary
Integrative pipelines combine TCGA genomics, proteomics and curated interaction databases to derive actionable models: a 14‑gene prognostic signature for ovarian cancer found by LASSO Cox and validated on independent protein cohorts; near‑exhaustive searches produce ~2,000 six‑gene prognostic signatures and a 9,515‑gene set linking to networks; drug repurposing via CMAP yields ~250 candidates with 23 retrospective NCI60 validations and 11 novel predictions, one top compound inhibiting lung cancer lines and reversing transcriptional changes; computational interaction prediction achieves 91% recall with 1% FDR and expands the human interactome to ~800,000 edges while flagging duplicated samples and missing clinical covariates.
AI-written summary · about
chapters (7)
- 0:00Start
- 4:50Cancer Data Integration Portal DIP
- 7:48A 3-Step Method for Model Selection
- 8:35Training & "Independent" Validation
- 11:47NSCLC: A Paradigm Shift
- 22:10Validating Predictions
- 25:59Conclusions
related talks
30mUnlocking the Genome: Exploring the Potential of CRISPR/Cas9
27mPractical Quantum Computing
26mTed Willke, Senior Principal Engineer, Intel Labs
21mEnabling the next level of scientific exploration
33mUnconventional Computing
27mTech History - An Outlook to the Future
27mComputers for Super-Human AI
18mIntersection of AI + Quantum Computing