Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Nov 19, 2014

• forgotten updates to feature selection

  iskandr

  committed
  71f082d

  Copy full SHA for 71f082d

Commits on Oct 30, 2014

• moved pairwise amino acid tables into pepdata repository

  iskandr

  committed
  1ab5f7d

  Copy full SHA for 1ab5f7d

• moved out of date predictors to 'old' directory, need to bring them back with selected features later

  iskandr

  committed
  3223f79

  Copy full SHA for 3223f79

Commits on Oct 29, 2014

• check column type in feature selector, drop columns with extremely small variance (even if it's nonzero)

  iskandr

  committed
  bb12099

  Copy full SHA for bb12099

• add neighboring MHC features and use rows of pairwise matrices to represent residues

  iskandr

  committed
  77df743

  Copy full SHA for 77df743

Commits on Oct 22, 2014

• drop features with zero variance

  iskandr

  committed
  ce2fc8d

  Copy full SHA for ce2fc8d

• cleaned up printing in feature selection

  iskandr

  committed
  54db9fc

  Copy full SHA for 54db9fc

Commits on Oct 21, 2014

• generalizing feature selection for use with HAI data

  iskandr

  committed
  f0a5f49

  Copy full SHA for f0a5f49

Commits on Oct 14, 2014

• added RandomForest to the collection of models

  iskandr

  committed
  1a4d3ee

  Copy full SHA for 1a4d3ee

• normalize by subtracting mean of each feature before training

  iskandr

  committed
  e9932fe

  Copy full SHA for e9932fe

• feature selection over >3k candidate features, both univariate and pairwise

  iskandr

  committed
  733ac6b

  Copy full SHA for 733ac6b

• generate lots of candidate features into HDF5 file

  iskandr

  committed
  9479646

  Copy full SHA for 9479646

• filtering aligned sequences

  iskandr

  committed
  66b4be0

  Copy full SHA for 66b4be0

Commits on Oct 13, 2014

• start pseudoseq extractor

  iskandr

  committed
  7e5d404

  Copy full SHA for 7e5d404

• make sure aligned strings are of correct (reference) length

  iskandr

  committed
  e252fe8

  Copy full SHA for e252fe8

• longer deletions in simple aligner

  iskandr

  committed
  ecdc7c5

  Copy full SHA for ecdc7c5

• write to output file

  iskandr

  committed
  4cb74d5

  Copy full SHA for 4cb74d5

• multi-species alignment via iterative deletion

  iskandr

  committed
  4c44256

  Copy full SHA for 4c44256

Commits on Sep 24, 2014

• removed bigger assay data

  iskandr

  committed
  4f5cd67

  Copy full SHA for 4f5cd67

• playing with sequence alignment

  iskandr

  committed
  7044a6e

  Copy full SHA for 7044a6e

Commits on Sep 15, 2014

• ensemble of 3 RandomForests trained on different output labels

  iskandr

  committed
  5cfd6e2

  Copy full SHA for 5cfd6e2

Commits on Sep 11, 2014

• removed unused files

  iskandr

  committed
  07442f0

  Copy full SHA for 07442f0

• using kmeans to make sub-models

  iskandr

  committed
  9e1c892

  Copy full SHA for 9e1c892

• exploring equivalence of assays

  iskandr

  committed
  6228cb4

  Copy full SHA for 6228cb4

• using ordinary amino acid features

  iskandr

  committed
  7b40543

  Copy full SHA for 7b40543

• plotting assay values against each other

  iskandr

  committed
  27305d2

  Copy full SHA for 27305d2

Commits on Sep 9, 2014

• RF doesn't make iteratively estimated features any better

  iskandr

  committed
  bd120c2

  Copy full SHA for bd120c2

• iterative trainingn with cleaner data

  iskandr

  committed
  ef7d679

  Copy full SHA for ef7d679

• clean your data fools

  iskandr

  committed
  dacbb05

  Copy full SHA for dacbb05

Commits on Sep 8, 2014

• making a more inclusive IEDB dump in mhc.csv, storing group epitope data in mhc_grouped.csv

  iskandr

  committed
  27c3349

  Copy full SHA for 27c3349

• download IEDB data directly, generated broader dataset

  iskandr

  committed
  b4a8607

  Copy full SHA for b4a8607

Commits on Sep 5, 2014

• experimenting with auto-selection of regularization params

  iskandr

  committed
  654619b

  Copy full SHA for 654619b

Commits on Sep 4, 2014

• using survival predictors

  iskandr

  committed
  d0cba54

  Copy full SHA for d0cba54

Commits on Aug 30, 2014

• refactored, leave-one-out CV for alleles

  iskandr

  committed
  3779c2e

  Copy full SHA for 3779c2e

Commits on Jul 17, 2014

• use two-pass regression on last iteration of feature matrix loop

  iskandr

  committed
  84629d9

  Copy full SHA for 84629d9

Pagination

Previous

Next