pyIsoPEP 0.1.0

Estimate PEPs from empirical null models using isotonic regression.

Isotonic PEP Estimation

Overview

pyIsotonicPEP provides a unified interface for estimating Posterior Error Probabilities (PEPs) using isotonic regression for identifications in shotgun proteomics. It supports two methods:

• q2pep: Estimate non-decreasing PEPs from q-values.
• d2pep: Estimate non-decreasing PEPs from a stream of target and decoy observations derived from target-decoy competition (TDC) method.

The package consists of two main Python files:

• IsotonicPEP.py: Implements isotonic regression for PEP estimation.
• main.py: Provides a command-line interface (CLI) to run the isotonic regressor with various options.

Features

• Isotonic Regression: Enforces a non-decreasing constraint on probability estimates.
• q2pep Method: Converts q-values to PEP values.
• d2pep Method: Processes target-decoy observations to compute PEP values.
• Optional PEP-based q-value Estimation: Calculates q-values from the estimated PEPs.
• Flexible Input Formats: In d2pep mode, supports both concatenated input files and separate target/decoy files.
• Configurable Regression Options: Choose between PAVA and I-Spline regression, with options for interpolation and block center calculation.

Usage

Pull the Docker image from GitHub Container Registry and display the help message

$ podman pull ghcr.io/statisticalbiotechnology/pyisotonicpep:main
$ podman run --rm -it pyisotonicpep:main -h

Command-Line Options

General Options

usage: main.py [-h] [--no-calc-q] [--verbose] {q2pep,d2pep} ...

• {q2pep,d2pep} Select the PEP estimation method.
  • q2pep Estimate PEPs from q-values.
  • d2pep Estimate PEPs from target-decoy observations.
• -h, --help Show this help message and exit.
• --no-calc-q Do not estimate q-values from calculated PEPs (by default, q-values are calculated).
• --verbose Print detailed parameter information.

q2pep Mode Options

usage: main.py q2pep [-h] --input INPUT [--qcol QCOL] [--regression-algo {PAVA,ispline}] [--max-iter MAX_ITER] [--ip] [--ip-algo {ispline,pchip}] [--center-method {mean,median}] --output OUTPUT

• --input INPUT Path to the TSV file containing q-values.
• --qcol QCOL Column name for q-values in the input file (default: 'q-value').
• --regression-algo {PAVA,ispline} Regression algorithm to use (default: 'ispline').
• --max-iter MAX_ITER Maximum iterations for I-Spline iterative solvers (default: 5000).
• --ip Apply monotonic interpolation (only used when using the PAVA regression algorithm).
• --ip-algo {ispline,pchip} Interpolation algorithm for PAVA-derived block centers (default: 'ispline').
• --center-method {mean,median} Method for computing block centers in interpolation (default: 'mean').
• --output OUTPUT Output file path or directory. If a directory is provided, the default filename outputPEP.target.qbased.txt will be used.

d2pep Mode Options

usage: main.py d2pep [-h] (--cat-file CAT_FILE | --target-file TARGET_FILE) [--decoy-file DECOY_FILE] [--score-col SCORE_COL] [--type-col TYPE_COL] [--target-label TARGET_LABEL] [--decoy-label DECOY_LABEL] [--regression-algo {PAVA,ispline}] [--max-iter MAX_ITER] --output OUTPUT

• --cat-file CAT_FILE Path to a concatenated TSV file containing score and label columns.
• --target-file TARGET_FILE Path to the TSV file containing target scores (used in separate input mode).
• --decoy-file DECOY_FILE Path to the TSV file containing decoy scores (required in separate input mode).
• --score-col SCORE_COL Column name for score (default: 'score').
• --type-col TYPE_COL Column name for target/decoy (default: 'label').
• --target-label TARGET_LABEL Target label in concatenated input file (default: 'target').
• --decoy-label DECOY_LABEL Decoy label in concatenated input file (default: 'decoy').
• --regression-algo {PAVA,ispline} Regression algorithm to use (default: 'ispline').
• --max-iter MAX_ITER Maximum iterations for I-Spline iterative solvers (default: 5000).
• --output OUTPUT Output file path or directory. If a directory is provided, the default filename outputPEP.target.dbased.txt will be used (note: only target PEPs are saved).

Running examples with the provided files

1. q2pep Mode

$ podman run --rm -it -v .:/data pyisotonicpep:main q2pep --input /example/peptide.target.txt --output /data

2. d2pep Mode

2.1 using a concatenated target and decoy input

$ podman run --rm -it -v .:/data pyisotonicpep:main d2pep --cat-file /example/peptide.cat.txt --score-col score --type-col type --target-label 0 --decoy-label 1 --output /data

2.2 using separate target and decoy inputs

$ podman run --rm -it -v .:/data pyisotonicpep:main d2pep --target-file /example/peptide.target.txt --decoy-file /example/peptide.decoy.txt --score-col score --output /data