Gene Extrapolation Models for Toxicogenomic Data

545 views

Published on

Published in: Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
545
On SlideShare
0
From Embeds
0
Number of Embeds
10
Actions
Shares
0
Downloads
10
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Gene Extrapolation Models for Toxicogenomic Data

  1. 1. geneEXTRAPOLATIONmodels forTOXICOGENOMICdata daniel gusenleitner nacho caballero
  2. 2. Testing for carcinogenicityis costly
  3. 3. Genes show clustered responsesExpression correlates between platforms
  4. 4. We want to extrapolate the expression of regular genes 11000 Genes2K Arrays 10K 1K Regular Landmark Genes Genes
  5. 5. We fit a linearmodel to each 2K Arrays 1Kregular gene X Landmark GenesPredicted Expression = XβExpression Gene 1 = X1β1 +X2β2 +…+X2Kβ2KExpression Gene 2 = X1β1 +X2β2 +…+X2Kβ2K …Expression Gene 10K = X1β1 +X2β2 +…+X2Kβ2K
  6. 6. ElasticNet mean error number of variables glmnet: Lasso and elastic- net regularized generalized linear models http://cran.r-project.org/web/ packages/glmnet/index.html
  7. 7. Neural Networks regular genes hidden layernnet: Feed-forwardNeural Networks andMultinomial Log-Linear Models landmarkhttp://cran.r-project.org/ genesweb/packages/nnet/index.html
  8. 8. mean fluorescent intensitysignal ratio Intensity variation -to-noise intensity standard deviationSNR = extrapolation mean error
  9. 9. Building 10451 models takes a long time… runtime single total per CPU runtime model runtime linear 120 x 3 h 2 min 360 hregression elastic 120 x 16 h 11 min 1920 h net neural 50 x 0.75 7800 h 45 min network h ?
  10. 10. Signal-to-Noise Comparison E-net LM NN ENSRNOG00000013133 135.58 19.62 13.21 ENSRNOG00000011861 209.82 28.82 12.08 ENSRNOG00000033466 190.81 26.58 11.86 ENSRNOG00000036816 197.82 23.09 9.93 ENSRNOG00000003515 273.62 29.35 8.68 ENSRNOG00000002254 53.43 8.83 7.21 ENSRNOG00000031266 76.19 8.19 6.70 ENSRNOG00000005963 145.06 6.99 6.49 ENSRNOG00000008613 38.86 3.97 6.07 ENSRNOG00000023095 13.57 2.70 5.98 ENSRNOG00000020947 17.27 2.41 5.04 ENSRNOG00000007258 103.77 13.71 4.91 ENSRNOG00000019813 16.53 3.01 4.68 ENSRNOG00000014232 61.69 9.17 4.05 ENSRNOG00000002454 50.71 5.58 3.80 ENSRNOG00000018201 5.04 1.64 3.39
  11. 11. The elastic net outperformsstandard linear regression Signal-to-noise ratio Elastic Linear Net Regression
  12. 12. Additional feature selectionPerformance of extrapolation models on carcinogenicity classifiersCorrelation between Luminex and Affymetrix chips

×