wsjung / netprophet_3.0

NetProphet 3.0

NetProphet3.0 is a package for network inference: first, it generates intermediate networks by regression and other algorithms LASSO, DE, BART, PWM. Then, it combines them into one final network using XGboost, a supervised ML. The easiest way to run NetProphet3 is with the singularity container, as demonstrated below. Steps are easy: first install the singularity software, then pull the container, you will be ready to run NetProphet3.

I. Install NetProphet 3.0

The easiest, with Singularity container

• Refer to Singularity website and install singularity >=3.5.2
• load the s_np container from the sylab cloud with this command

  singularity pull library://dabid/default/s_np3:sha256.e14c3a078c052932b5e2a840beea1fa5b8798db5668e196d8e297b223b6390d3
  

• rename the container with a shorted file name, so it is easier to use.

  mv s_np3_sha256.e14c3a078c052932b5e2a840beea1fa5b8798db5668e196d8e297b223b6390d3.sif s_np3.sif
  

The more advanced, here, all dependencies have to be installed

Refer to this wiki page for installing NetProphet dependencies

II. Run NetProphet 3.0

Clone NetProphet3.0 in a path in your computer, let's call that path ${p_src_code}

With a toy example

This is an example of NetProphet3.0 combination command, it assumes that the singularity container will be used and that's why --flag_singularity ON. Refer to the folder ${p_src_code}toy_example, for commands of each of the NetProphet3 modules; they can be run with and without singularity.

p_wd=/scratch/mblab/dabid/proj_net/
p_src_code=${p_wd}code/NetProphet_3.0/
p_out_dir=${p_wd}code/NetProphet_3.0/toy_example/res/

${p_src_code}np3 -c \
    --p_in_binding_event ${p_src_code}toy_example/data_binding_reg_target.tsv \
    --l_in_name_net "lasso,de,bart,pwm" \
    --l_in_path_net "${p_out_dir}features/net_lasso.tsv,${p_src_code}toy_example/data_zev_de_shrunken_50_500_indexed,${p_out_dir}features/net_bart.tsv,${p_out_dir}features/net_pwm.tsv" \
    --flag_training ON-CV \
    --combine_cv_nbr_fold 10 \
    --p_out_dir ${p_out_dir}10cv/ \
    --flag_singularity ON \
    --p_singularity_img /path/singularity/image \
    --p_singularity_bindpath /your/home/dir/ \
    --flag_slurm OFF

More about np3 command and options

• Help usage: ./${p_src_code}np3 -h
• NetProphet3.0: with SLURM environment
• NetProphet3.0: with Singularity
• NetProphet3.0: combination method
  • Train for 10CV
  • Train for integration
  • No Training, use a pre-built model (yeast)
• NetProphet3.0: construction of intermediate networks
  • LASSO
  • BART
  • PWM