| 1 |
Example call of build_splitsets.rb |
| 2 |
./build_splitsets.rb -i ./clustered_sequences/sprot_go_original.f -s 80 -o ./clustered_sequences/cv_whatever/ -f 10 -p |
| 3 |
|
| 4 |
Builds the training, testing and holdoutsets out of the sequencefile given after the -i parameter (./clustered_sequences/sprot_go_original.f). |
| 5 |
The test/training set for a 10 fold (-f parameter) crossvalidation will contain 80% (-s parameter) of the sequences contained in ./clustered_sequences/sprot_go_original.f. |
| 6 |
The -p options aktivates the plotting of the distributions of the GO anntations within ./clustered_sequences/sprot_go_original.f. -p is optional. |
| 7 |
All outputfiles are put into ./clustered_sequences/cv_whatever/ (-o parameter) |
| 8 |
Files contained in the folder are: |
| 9 |
cv_split_80_i i = 1-10 # The 10 sets for crossvalidation |
| 10 |
holdout80_set.f # holdout set containing 20% of the sequences from sprot_go_original.f |
| 11 |
crossvalidation_combinatorics.txt # tabular file where each ,ine represents one combination of the 10 crossvalidation sets (cv_split_80_i). |
| 12 |
# The last file is always the testset! |
| 13 |
Rplots80.pdf # Plots for the distributions |