This script (i) splits combined RNA-Seq & hyperspectral data for each transcript into train and test datasets, (ii) runs variable selection to find the lowest MSE, and trained/calibrated using gene expression as response variable and spectral data of selected bands as explanatory variable.
-
-c, --transcript_csv
- CSV for a single transcript containing all spectra
-
-pod, --plot_out_dir
- Plot/figure output directory
-
-cod, --csv_out_dir
- CSV output directory (calibration, test, and permutation scores)
-
-o, --outdir
- Main/root output directory
-
-m, --model_out_dir
- PLSR model sav file output directory
-
-p, --permutation_out_dir
- Permutation numpy array output directory
-
-t, --test_size
- Test size for train/test split (default=0.25)
To run the scrip, you have to specify at least the CSV:
./rnaseq_spectral_plsr.py -c ./rnaseq_spectra/Gh_A01G026200_tpm_spectra.csv