GRN with Gradient Boosting

Utilities and runner scripts for generating gene regulatory networks with arboreto, XGBoost and Light GBM.

Usage

Runner can be used to run either XG Boost or light GBM using the AnnData h5ad file or CSV file as input

Command-line parameters to select method, statistics and output files

Options to provide device, method, statistics output file and network output file are mandatory irrespective of wether the input is a AnnData h5ad file or CSV file. The options are provided as follows:

python -m grb.cli
        [-h] [--take_n TAKE_N]
        [--device {cpu,gpu}]
        [--method {xgb,sgbr,arb:default,arb:gbm,lgb}]
        [--use_tqdm]
        [--rstats_out_file RSTATS_OUT_FILE]
        [--out_file OUT_FILE]
        {csv,ad} ...

Generate GRNs w. XGB/lightgbm for Single Cell Data

positional arguments:
  {csv,ad}              CSV or AnnData H5AD file
    csv                 CSV File as Input
    ad                  H5AD File as Input

options:
  -h, --help            show this help message and exit
  --take_n TAKE_N
  --device, -c {cpu,gpu}
  --method, -m {xgb,sgbr,arb:default,arb:gbm,lgb}
  --use_tqdm
  --rstats_out_file RSTATS_OUT_FILE
  --out_file OUT_FILE

Command-line parameters for csv file as input

Option to provide csv file as input is as follows:

 python -m grb.cli csv
        [-h]
        --csv_file CSV_FILE

 options:
   -h, --help           show this help message and exit
   --csv_file CSV_FILE

Command-line parameters for anndata h5ad file as input

Option to provide AnnData h5ad file as input is as follows:

python -m grb.cli ad
       [-h]
       [--ntop_genes NTOP_GENES]
       [--data {dense,sparse}]
       [--nsub_cells NSUB_CELLS]
       [--tf_file TF_FILE]
       [--h5ad_file H5AD_FILE]
       [--select_hvg SELECT_HVG]

options:
  -h, --help            show this help message and exit
  --ntop_genes NTOP_GENES
  --data, -d {dense,sparse}
  --nsub_cells NSUB_CELLS
  --tf_file TF_FILE
  --h5ad_file H5AD_FILE
  --select_hvg SELECT_HVG