PPTStab

Designing of thermostable proteins with a desired melting temperature

Introduction

PPTStab is developed to predict the thermostability of proteins and design thermostable proteins. The standalone version uses an ANN+MLP ensemble regressor model. PPTStab is also available as a web-server at https://webs.iiitd.edu.in/raghava/pptstab. Please read/cite the content about PPTStab for complete information including the algorithm behind the approach.

Installation

To use PPTStab, follow these steps to set up your environment:

Install using environment.yml

1. Create a new Conda environment from the environment.yml file, replacing <env_name> with your preferred name (e.g. pptstab):

conda env create -n <env_name> -f environment.yml

Example:

conda env create -n pptstab -f environment.yml

Note: If an environment with that name already exists, remove it first:

conda env remove -n <env_name>
conda env create -n <env_name> -f environment.yml

2. Activate the environment:

conda activate pptstab

Required Dependencies

The standalone version of PPTStab is written in Python 3.11. The following libraries are required:

Package	Version
numpy	1.26.4
scikit-learn	1.6.1
tensorflow	2.16.1
torch	2.3.1
transformers	4.41.2
pandas	2.3.3
onnxruntime	1.18.0
joblib	1.4.2
tqdm	4.66.4

Minimum Usage

To see all available options, type:

python pptstab.py -h

To run the example (predict melting temperature using default settings):

python pptstab.py -i example.fasta -f 1

Results will be saved to outfile.csv by default.

Full Usage

usage: pptstab.py [-h] [-i INPUT] [-o OUTPUT] [-j {1,2}] [-d {1,2}] [-f {0,1}] [-m {EMB,AAC,SER}]

Arguments

Argument	Description	Default
-i INPUT	Input protein/peptide sequence(s) in FASTA format or one sequence per line	Required
-o OUTPUT	Output file name	outfile.csv
-j {1,2}	Job type: 1 = Predict, 2 = Design	1
-f {0,1}	Flag: 1 = lysate, 0 = cell	1
-d {1,2}	Display: 1 = thermophilic proteins only, 2 = all proteins	1
-m {EMB,AAC,SER}	Method: EMB (ProtBERT embeddings), AAC (amino acid composition), SER (Shannon entropy)	EMB

Usage Examples

Predict Tm using default ProtBERT embeddings:

python pptstab.py -i example.fasta -f 1 -m EMB

Predict Tm using SER method (lysate context):

python pptstab.py -i example.fasta -f 1

Predict Tm using AAC method, cell context, show all results:

python pptstab.py -i example.fasta -f 0 -m AAC -d 2

Design thermostable mutants using SER method:

python pptstab.py -i example.fasta -j 2 -f 1 -m SER -o design_output.csv

Input File: Accepts sequences in FASTA format or one sequence per line (single-letter amino acid code).

Output File: Results saved in CSV format. If no output file is specified, results are stored in outfile.csv.

Job: 1 = Predict melting temperature (Tm) for input sequences. 2 = Design mode — generates all single-point mutants of input sequences and predicts Tm for each.

Flag: Sets the experimental context. 1 = lysate (cell-free system), 0 = cell. Default is 1 (lysate).

Method: Selects the feature/model type. SER uses Shannon entropy per residue, AAC uses amino acid composition, EMB uses ProtBERT protein embeddings (requires GPU for large inputs).

PPTStab Package Files

File	Description
INSTALLATION	Installation instructions
LICENSE	License information
README.md	This file
pptstab.py	Main Python program
environment.yml	Conda environment file
example.fasta	Example input file with protein sequences in FASTA format
example_predict_output.csv	Example output for predict module
example_design_output.csv	Example output for design module
models/	Pre-trained model files (ANN ONNX + MLP pkl) for each method