======================================================= PlifePred2: Peptide Half-Life Prediction & Design Suite

PlifePred2 is a comprehensive toolkit for peptide half-life prediction and rational design using integrated machine learning, sequence-derived descriptors, and physicochemical analysis.

It supports:

• Standalone half-life prediction
• Exhaustive single mutation scanning (design mode)

The toolkit is optimized for Linux and macOS environments and supports reproducible conda-based installation.

.. image:: https://img.shields.io/badge/python-3.10%2B-blue.svg :target: https://www.python.org/

.. image:: https://img.shields.io/badge/license-GPLv3-green.svg :target: https://www.gnu.org/licenses/gpl-3.0

Overview

PlifePred2 predicts peptide half-life using two complementary models:

• Model 1 — Natural peptides
• Model 2 — Modified peptides (modification flags)

It provides two workflows:

1. Prediction Mode — Estimate half-life of input peptides
2. Design Mode — Generate all single mutants and rank them by predicted stability

Core Functionalities

• Peptide half-life prediction
• Physicochemical property calculation (pI, MW, charge, etc.)
• Exhaustive single mutation scanning
• Automatic sequence filtering and validation

Installation

Option 1 — Conda Environment (Recommended)

.. code-block:: bash

conda env create -f environment.yml git clone https://github.com/raghavagps/plifepred2.git cd plifepred2 wget https://github.com/patrik-ackerman/plifepred2/releases/download/v.0.1/pfeature_comp chmod +x pfeature_comp

Option 2 — Conda environment from scratch:

.. code-block:: bash

conda create -n plifepred2 python=3.10 conda activate plifepred2 conda install pandas numpy scikit-learn==1.4.2 joblib

Clone repository:

.. code-block:: bash

git clone https://github.com/raghavagps/plifepred2.git cd plifepred2 wget https://github.com/patrik-ackerman/plifepred2/releases/download/v.0.1/pfeature_comp chmod +x pfeature_comp

Usage Overview

PlifePred2 provides two main scripts:

• plifepred2.py — Prediction mode
• design.py — Mutation design mode

Prediction Module

General usage:

.. code-block:: bash

python plifepred2.py -i input.fasta -m -o output.csv

Arguments

+------------+--------------------------------------------+ | Argument | Description | +============+============================================+ | -i | Input multi-FASTA file | +------------+--------------------------------------------+ | -m | Model type (1 or 2) | +------------+--------------------------------------------+ | -o | Output CSV file | +------------+--------------------------------------------+ | -f | Modification flags (Model 2 only) | +------------+--------------------------------------------+ | -p | Physicochemical properties (optional) | +------------+--------------------------------------------+

Model 1 — Natural Peptides

.. code-block:: bash

python plifepred2.py -i input.fasta -m 1 -o output.csv

Model 2 — Modified Peptides

Modification flags:

+------+------------------------------+ | Flag | Description | +======+==============================+ | 0 | D-amino acid | +------+------------------------------+ | 1 | C-terminal modification | +------+------------------------------+ | 2 | N-terminal modification | +------+------------------------------+ | 3 | Cyclization | +------+------------------------------+ | 4 | Post-translational mod | +------+------------------------------+

Example:

.. code-block:: bash

python plifepred2.py -i input.fasta -m 2 -f 1,3 -o output.csv

Optional Physicochemical Properties

Use -p to compute properties.

Available properties:

+------+----------------------+ | Code | Property | +======+======================+ | 1 | Hydrophobicity | +------+----------------------+ | 2 | Steric hindrance | +------+----------------------+ | 3 | Hydropathicity | +------+----------------------+ | 4 | Amphipathicity | +------+----------------------+ | 5 | Hydrophilicity | +------+----------------------+ | 6 | Net hydrogen | +------+----------------------+ | 7 | Net charge | +------+----------------------+ | 8 | Isoelectric point | +------+----------------------+ | 9 | Molecular weight | +------+----------------------+

Examples:

.. code-block:: bash

python plifepred2.py -i input.fasta -m 2 -f 1 -p 8,9 -o output.csv python plifepred2.py -i input.fasta -m 1 -p 8,9 -o output.csv

Output Format

+-----------+--------------------------------------+ | Column | Description | +===========+======================================+ | ID | Sequence identifier | +-----------+--------------------------------------+ | Sequence | Peptide sequence | +-----------+--------------------------------------+ | Halflife | Predicted probability | +-----------+--------------------------------------+ | [Props] | Optional selected properties | +-----------+--------------------------------------+

Design Module

The design module performs exhaustive single mutation scanning.

General usage:

.. code-block:: bash

python plifepred2_design.py -i input.fasta -o design_output.tsv

For modified peptides:

.. code-block:: bash

python plifepred2_design.py -i input.fasta -f 1 -o design_output.tsv

Design Output Format

+------------+----------------+-----------+-------+ | Seq_ID | Mutant_ID | Sequence | Score | +============+================+===========+=======+ | Example | A5V | ACDVFG... | 0.87 | +------------+----------------+-----------+-------+

Mutants are sorted from highest to lowest predicted stability.

Sequence Validation

• Only standard amino acids allowed
• Length restriction: 12–100 residues
• Invalid sequences logged separately

Machine Learning Model

Algorithm: Random Forest Regressor

Training version: scikit-learn 1.4.2

If you encounter version mismatch warnings:

.. code-block:: bash

pip install scikit-learn==1.4.2

Citation

If you use PlifePred2, please cite the corresponding publication.

Support

GitHub: https://github.com/raghavagps/plifepred2

Email: raghava@iiitd.ac.in

License

GPLv3 License

This software is distributed under the GNU General Public License v3.0.