Accelerating drug research with AI and cloud computing
summary
Budget-conscious RNA‑seq planning recommends using paired design to reduce inter‑individual variation and absorb batch effects, at the cost of halving degrees of freedom. Empirical example: COVID project with 18 patients and 79 samples yields 76 vs 2 DE genes in blood when paired versus unpaired (nasal tissue showed 40 paired vs 51 unpaired). Model counts as negative binomial/gamma‑Poisson, use batch‑correction tools like SVA and RUVseq and validate hits with PantherDB. Costs typically start >$500 per sample.
AI-written summary · about
chapters (21)
- 0:00Intro
- 2:08Why do you need a statistician
- 5:50How many samples do you need
- 12:48RNA characterization
- 15:07Approach
- 15:45Methods
- 16:42Batch effects
- 17:25Pair design
- 17:55Meaningful metadata
- 18:33Filter genes
- 18:54Results
- 20:52Molecular Functions
- 22:29Reducing individual variation
- 23:02Pairing absorbs batch effects
- 23:45Technical downsides
- 24:14Medical implication
- 24:56Utility
- 25:46Examples
- 26:34ALS
- 27:28Genomics
- 28:27Thanks
related talks
14mBig data, AI, the genome, and everything
42mHow Novel Compute Technology Transforms Life Science Research
10mGenesis Final Round - Drylab AI
17mBig data in personalized medicine
21mFighting Cancer with AI
45mDigital Disruption in health & medical research
23mGenomics and AI on Cloud Foundry
21mOpen Biology: Revolutionizing biological research