PPIprophet: Protein-Protein Interaction Prediction Suite

PPIprophet is a Python tool for predicting protein-protein interactions (PPIs) from proteomics data. It supports a variety of quantitation schemes and provides results compatible with Cytoscape for network analysis. This README covers installation, input formats, parameter configuration, running the tool, and interpreting results.

---

Table of Contents

• PPIprophet: Protein-Protein Interaction Prediction Suite
  • Table of Contents
  • Overview
  • Installation
  • Input File Format
  • Experimental Information File
  • Parameter Configuration
  • Running PPIprophet
  • Output Files
  • Importing Results into Cytoscape
  • Troubleshooting
  • Contact

---

Overview

PPIprophet predicts protein-protein interactions from proteomics data using deep learning and network analysis. It is highly configurable and supports various input formats and quantitation schemes (MS1/MS2 XIC, spectral counts, TMT, SILAC, etc.).

PPIprophet is designed for co-fractionation MS experiments and large-scale PPI prediction.

---

Installation

1. Clone the repository and navigate to the project folder.
2. Install dependencies:

   pip install -r requirements.txt

---

Input File Format

Input files must be wide-format matrices with these requirements:

• First column: GN (Gene name or protein ID, unique for each row)
• Remaining columns: Quantitation values, ordered by fractionation scheme (column names are flexible)

Note: Duplicate rows in the GN column will trigger an error.

Example:

GN	Fraction1	Fraction2	...
A	20400	...
D	1230120	...
C	1230120	...

All common proteomics quantitation schemes are supported. Example files are provided in test/test_fract.txt.

---

Experimental Information File

The sample IDs file (e.g., sample_ids.txt) must have the following columns:

Sample	cond	group	short_id	repl	fr
./Input/c1r1.txt	Ctrl	1	ipsc_2i_1	1	65
./Input/c1r2.txt	Ctrl	1	ipsc_2i_2	2	64
./Input/c1r3.txt	Ctrl	1	ipsc_2i_3	3	65
./Input/c2r1.txt	Treat1	2	ipsc_ra_1	1	65
./Input/c2r2.txt	Treat1	2	ipsc_ra_2	2	65
./Input/c2r3.txt	Treat1	2	ipsc_ra_3	3	65

• Sample: Full path to the file to process (must match the actual filename)
• cond: Condition name (Ctrl, Treat1, Treat2, etc.)
• group: Group number (1 for control)
• short_id: Alternative ID
• repl: Replicate number
• fr: Number of fractions per file

Note:

• The Sample column must match the input file names exactly (including extension).
• For multiple conditions, use Ctrl, Treat1, Treat2, etc. as labels.

---

Parameter Configuration

Parameters can be set via the ProphetConfig.conf file or directly on the command line. Command-line parameters override the config file and are saved to it.

Argument	Description	Default
-db	Path to PPI network file (STRING format)	None
-fdr	Global FDR threshold	0.3
-sid	Path to sample IDs file	sample_ids.txt
-out	Output folder name	Output
-crapome	Crapome contaminant reference file (optional)	crapome.org.txt
-thresh	Frequency threshold for Crapome filtering	0.5
-skip	Skip preprocessing/feature generation (True/False)	False

To see all available parameters:

python3 main.py --help

---

Running PPIprophet

After installing dependencies, you can test PPIprophet with the example dataset:

python3 main.py -sid test/test_ids.txt

To run with all default settings:

python3 main.py

To specify parameters, use command-line arguments. For example, to use a custom interaction network and a 10% FDR:

python3 main.py -db myppi.txt -fdr 0.1

Running PPIprophet with Docker

Recommended

# start in ppiprophet directory
cd [path_to]/PPIprophet

# build docker image
docker build ./ -t ppiprophet

# launch docker container from PPIprophet directory and enter a bash shell
docker run --rm -it -v`pwd`:/wd -w /wd  ppiprophet bash

# see PPIprophet options
python main.py --help

# launch PPIprophet using files in cwd
python  main.py -sid test_data/sample_ids.txt

# output appears in ./Output
less Output/*

# exit container
exit

---

Output Files

PPIprophet generates two main folders:

• tmp/: Intermediate files for debugging/validation (can be deleted after completion)
• Output/: Final results, including:
  • adj_list.txt: PPI list (proteinA/proteinB/Probability) and Crapome frequencies
  • communities.txt: Modules detected after clustering
  • d_scores.txt: Interaction probabilities as modified WD scores
  • probtot.txt: Adjacency list with filtered interactions (FDR applied)
  • prot_centr.txt: Protein-centric interactors list

File names may vary slightly depending on the run (see output folder for details).

---

Importing Results into Cytoscape

• probtot.txt and d_scores.txt can be imported directly into Cytoscape.
• adj_list.txt can be used to visualize interaction frequencies.

---

Troubleshooting

Common errors and solutions:

• NaRowError / NaInMatrixError: 'NA' values in input; replace with 0.
• MissingColumnError: Required columns (GN or ID) missing.
• DuplicateRowError / DuplicateIdentifierError: Duplicate entries in GN column; ensure uniqueness.
• EmptyColumnError: A column contains only 'NA'.

Tip: If files have different numbers of fractions, add zero-filled columns as needed.

---

Contact

For questions or support, see the project README on GitHub.