3. reactIDR

reactIDR computation

Training

python ../reactIDR/IDR_hmm.py --train  --time 10 --core 6 --grid --output_param sample_parameters.txt --ref ref.fa --output sample.csv \
--case sample_case.ctss_bed.tab --cont sample_case.ctss_bed.tab

Trained parameters are written to sample_parameters.txt. Posterior probability is written to train_sample.csv.

cd example
python ../reactIDR/IDR_hmm.py training.argv

Options can be set using a recipe as well.

Test

python ../reactIDR/IDR_hmm.py --test   --time 10 --core 6        --param sample_parameters.txt --output fit_sample_parameters.txt --ref ref.fa --output sample.csv \
--case sample_case.ctss_bed.tab --cont sample_case.ctss_bed.tab

Posterior probability is written to test_sample.csv.

cd example
python ../reactIDR/IDR_hmm.py test.argv

Options can be set using a recipe as well.

Fit (=test and optimize furthermore)

python ../reactIDR/IDR_hmm.py --fit   --time 10 --core 6        --param sample_parameters.txt --ref ref.fa --output sample.csv \
--case sample_case.ctss_bed.tab --cont sample_case.ctss_bed.tab

Posterior probability is written to fit_sample.csv. Read initial parameters from sample_parameters.txt and then fit them again.

IDR computation

python ../reactIDR/IDR_hmm.py --global  --time 10 --core 6 --grid --param default_parameters.txt --ref ref.fa --output sample.csv \
--case sample_case.ctss_bed.tab --cont sample_case.ctss_bed.tab

Posterior probability is written to global_sample.csv. Without --idr option, 1-irreproducible discovery rate is written.

Option list

• idr (default: false)
  • output 1-probability
• time
  • iteration times
• core
  • multi-core computation
• ref
  • reference structure
• DMS
  • requires a fasta file to recognize ACGT
• output
  • suffix of the output file
• independent
  • independent optimization for each transcript