Statistical Genomics Lecture 21 Farm CPU Zhiwu Zhang
- Slides: 24
Statistical Genomics Lecture 21: Farm. CPU Zhiwu Zhang Washington State University
Administration Homework 5, due April 13, Wednesday, 3: 10 PM Final exam: May 3, 120 minutes (3: 10 -5: 10 PM), 50 Department seminar (April 4) , Nural Amin
Outline History of method and software development Farm. CPU BLINK
Problems in GWAS Computing difficulties: millions of markers, individuals, and traits False positives, ex: “Amgen scientists tried to replicate 53 high-profile cancer research findings, but could only replicate 6”, Nature, 2012, 483: 531 False negatives
Q EMMA PC PC+K EMMAx Q+K GWAS Stream MLMM P 3 D CMLM SELECT GCTA ECMLM FST-LMM GEMMA Gen. Abel Farm. CPU BLINK
Computing speed t test GLM Speed improvement Power improvement Gen. ABEL Fa. ST-LMM GEMMA P 3 D/EMMAX EMMA CMLM ECMLM Select SUPER MLMM MLM Power | type I error
Usage of Software Packages Software Leading Authors Corresponding authors PUMA Gabriel E. Hoffman Jason G. Mezey TATES Sophie van der Sluis GAPIT Lipka AE Zhang Z MLMM Vincent S C++ 2013 8 Fortran 2013 20 R 2012 106 R/python 2012 69 C++ 2012 88 C++ 2011 104 Fortran 2004 141 Sabatti C & Eskin E C++ 2010 349 Jian Y C++ 2011 380 R 2007 510 Java 2006 660 C++ 2007 7037 75% Nordborg M GEMMA Zhou X Stephens M Christoph L, Listgarten J, Fast. LMM Heckerman D Qxpak M. Pérez-Enciso EMMAX Kang HM GCTA Jian Y Language Released Citation M. Pérez-Enciso Gen. ABEL Aulchenko YS Bradbury PJ, Zhang Z, Kroon TASSEL Bradbury PJ DE PLINK Purcell S
Magnus Norborg Test WO correction Correction with MLM Nature 2010 GWAS does not work for traits associated with structure
Why human geneticists not go beyond PLINK?
MLM was more enriched on Flowering time genes
Model Development Si: Testing marker Q: Population structure Adjustment on marker K: Kinship S: Pseudo QTNs Adjustment on covariates
y = PC + Kinship + e SUPER algorithm y = PC + SNP + e -2 LL Bins QTNs y = PC + Kinship + SNP + e
y = PC + Kinship + e Farm. CPU algorithm y = PC + SNP + e -2 LL Bins QTNs y = PC + QTNs + SNP + e
Computing speed t test GLM Speed improvement Power improvement Gen. ABEL BLINK Farm. CPU Fa. ST-LMM GEMMA P 3 D/EMMAX EMMA CMLM ECMLM Select SUPER MLMM MLM Power | type I error
FARM-CPU Fixed model y = M 1 + … + M t + mi + e p 1 … NA … pl Mt Pt 1 … Ptj … Ptk … Ptl Pt … … … … … M 2 P 21 … P 2 j … P 2 k … P 2 l P 2 M 1 P 11 … P 1 j … P 1 k … P 1 l P 1 m 1 … mj … mk … ml Optimization SNP Substitution (Fixed And Random Model Circuitous Probability Unification) Random model y = u + e with Var(u)∝SVD(M)
Re-analysis of Arabidopsis data Xiaolei Liu
Flowering time genes enriched
Associations on flowering time
It is time for human geneticists to move forward
Substitution makes difference
Converge fast
Farm. CPU is computing efficient Testing 60 K SNPs
Half million individuals, half million SNPs three days But, PINK new version is faster
Ladder for high hanging fruits GAPIT EMMA, EMMAx, GCAT, Gen. ABEL Structure, Eigenstrate PLIK GLM t, F, X 2… Uncorrelated or equally correlated MLM TASSEL GAPIT CMLM ECMLM
- Zhiwu zhang
- 01:640:244 lecture notes - lecture 15: plat, idah, farad
- Harvest genomics
- Difference between structural and functional genomics
- Application of genomics
- A vision for the future of genomics research
- Genome is
- Essnet qsr
- "encoded genomics" -job
- Types of genomics
- Integrative genomics viewer
- Genomics
- Difference between structural and functional genomics
- "encoded genomics" -job
- Rachel butler bristol
- Functional genomics
- Integrative genomics viewer download
- Peza form 8105
- Physical farm planning
- Uses of farm structures and buildings
- Mollie animal farm represents
- Shunyuan zhang
- Jingxuan zhang
- Yunxi zhang
- Jing zhang rit