Insight Week 4

For 7/31/2015

Transcript

1. OncoPredictor Can we use genetic information to predict h ff ti d ill b f ti t? how effective a drug will be for a patient? Kyle Satterstrom
2. Data - Source Data Source • The Cancer Genome Atlas The Cancer Genome Atlas – Clinical and genetic records from thousands of cancer patients cancer patients I i d Data type # Features • I examined: – Disease: Glioblastoma Data type # Features Age / gender 2 Copy number 100 iRNA i 500 – Drug: Temozolomide – 94 patients total miRNA expression 500 mRNA expression 17,000 DNA methylation 24,000
3. Data - Groups 100% Data Groups 80% 100% rvivors 40% 60% Sur 20% 40% # d 0 500 1000 1500 2000 0% # days
4. Data - Groups 100% Data Groups Class I Class II 80% 100% rvivors 40% 60% Sur 20% 40% # d 0 500 1000 1500 2000 0% # days
5. Data - Features Data Features mRNA 6000 Methylation 15000 2000 4000 5000 10000 Log2 (Class II / Class I) -0.8 -0.4 0.0 0.4 0.8 0 Log2 (Class II / Class I) -0.1 0.0 0.1 0
6. Algorithm Algorithm • Random Forest Random Forest
7. Algorithm Algorithm • Random Forest Random Forest – 5-fold cross-validation: acy Accura Chance G ender C N V iRN A m RN A ylation d D ata Age / G e C m iR m R M ethyla C om bined D
8. Demo Demo • http://OncoPredictorme http://OncoPredictor.me
9. Feature importance Feature importance miRNA Feature Relative miRNA Feature Relative importance hsa-miR-9 6.1 hsa miR 572 4 9 hsa-miR-572 4.9 hsa-miR-301 4.8 hsa-miR-324-3p 4.2 hsa-miR-217 4.1 hsa-miR-202 4.0 hsa-miR-122a 3.8 hsa-miR-140 3.8 hsa-miR-651 3.7 h iR 598 3 6 hsa-miR-598 3.6
10. Feature importance Feature importance miRNA Feature Relative miRNA Feature Relative importance hsa-miR-9 6.1 hsa miR 572 4 9 hsa-miR-572 4.9 hsa-miR-301 4.8 hsa-miR-324-3p 4.2 hsa-miR-217 4.1 hsa-miR-202 4.0 hsa-miR-122a 3.8 hsa-miR-140 3.8 hsa-miR-651 3.7 h iR 598 3 6 hsa-miR-598 3.6
11. Feature importance Feature importance miRNA Feature Relative miRNA Feature Relative importance hsa-miR-9 6.1 hsa miR 572 4 9 hsa-miR-572 4.9 hsa-miR-301 4.8 hsa-miR-324-3p 4.2 hsa-miR-217 4.1 hsa-miR-202 4.0 hsa-miR-122a 3.8 hsa-miR-140 3.8 hsa-miR-651 3.7 h iR 598 3 6 hsa-miR-598 3.6
12. Feature importance Feature importance miRNA Feature Relative miRNA Feature Relative importance hsa-miR-9 6.1 hsa miR 572 4 9 hsa-miR-572 4.9 hsa-miR-301 4.8 hsa-miR-324-3p 4.2 hsa-miR-217 4.1 hsa-miR-202 4.0 hsa-miR-122a 3.8 hsa-miR-140 3.8 hsa-miR-651 3.7 h iR 598 3 6 hsa-miR-598 3.6
13. About me About me • Harvard Bioengineering PhD Harvard Bioengineering PhD – Longevity, delaying age-related disease H d C ll • Harvard College – Physics, History of Science