integration

Parameters

NetProphet can combine intermediate networks using integration method. TFs that have binding data, will be predicted by TFs-specific models, but TFs that do not have binding data, will be predicted by a model that was trained with all TFs.

-c for running the combination module.

--flag_training ON-IN for running the integration method.

--p_in_binding_event path of the positive binding labels, file is tab separated, header: REGULATOR TARGET.

--l_in_name_net a string of names of the intermediate networks separated by comma.

--l_in_path_neta string of paths of the intermediate networks separated by comma.

--in_nbr_reg the number of TFs that will be used to create TFs-specific models.

--p_out_dir path for output director, prediction and intermediate results will be save here.

Command

• These parameters are effective only if SLURM is used, flag_slurm ON:

  combine_slurm_mem memory allocated, it corresponds to the --mem parameter in SLURM (default 20G).

  combine_slurm_nodes number of nodes allocated, it corresponds to the --nodes parameter in SLURM (default 3).

  combine_slurm_ntasks number of tasks, it corresponds to the --ntasks parameter in SLURM (default 5).

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 \
    --in_nbr_reg 1 \
    --p_out_dir ${p_out_dir}integration/ \
    --flag_singularity ON \
    --p_singularity_img /path/singularity/image \
    --p_singularity_bindpath /your/home/dir/ \
    --flag_slurm ON \
    --mail_user xx@xx.edu \
    --mail_type FAIL \
    --p_out_dir_logs ${p_out_dir}log/ \
    --data zen \
    --combine_slurm_mem 10G \
    --combine_slurm_nodes 10 \
    --combine_slurm_ntasks 30