deeprm call

Run the inference workflow (prep → run → pileup). Heavy deps (pysam, pod5, torch) are optional and only needed for the relevant subcommands.

Group help

DeepRM Call (inference) Module

usage: deeprm inference [-h] {prep,run,pileup} ...

Positional Arguments

{prep,run,pileup}

Possible choices: prep, run, pileup

Sub-commands

prep

Preprocess raw inputs for inference

deeprm inference prep

run

Run model inference on preprocessed data

deeprm inference run

pileup

Aggregate predictions into site-level metrics

deeprm inference pileup

Examples

# 1) Preprocess inputs (external C++ binary or Python fallback)
deeprm call prep -i raw/ -o prep/ --threads 8

# 2) Run model inference
deeprm call run -m weight/deeprm_weights.pt -d prep/ -o pred/

# 3) Aggregate site-level metrics
deeprm call pileup -i pred/ -o pileup/ -b mpileup.filtered.pkl