CPPLocPred

Prediction of Cell-Penetrating Peptides and their Subcellular Localization

CPPLocPred is a two-stage machine learning tool:

• Stage 1 — Classifies peptides as CPP or Non-CPP using an ExtraTrees model (AAC features)
• Stage 2 — Predicts subcellular localization (Cytoplasm, Nucleus, Mitochondria, Endo/Lysosome, Others) using CatBoost models (DDR features)

It also supports motif scanning of protein sequences using MERCI for location-specific CPP motifs.

---

Requirements

python >= 3.6
pandas
scikit-learn
catboost
perl (for motif search only)

Install dependencies:

pip install pandas scikit-learn catboost

---

Directory Structure

CPPLocPred/
├── CPPLocPred.py                   # main script
├── MERCI_motif_locator.pl          # required for motif search (Job 2)
├── CPP_vs_NonCPP_ET_AAC.pkl
├── Cytoplasm_DDR_CatBoost.pkl
├── Nucleus_DDR_CatBoost.pkl
├── Mitochondria_DDR_CatBoost.pkl
├── Endo_lyso_DDR_CatBoost.pkl
├── Other_DDR_CatBoost.pkl
├── Localization_thresholds.pkl
└── motifs/
    ├── None/
    │   ├── Cytoplasm_motif
    │   ├── Nucleus_motif
    │   ├── Mitochondria_motif
    │   ├── Endo_lysosome_motif
    │   └── Others_motif
    ├── Koolman/          (same 5 files)
    ├── Betts-Russell/    (same 5 files)
    └── Rasmol/           (same 5 files)

---

Usage

python CPPLocPred.py -i INPUT -o OUTPUT [-j {1,2}] [options]

Flag	Description	Default
-i	Input FASTA file	required
-o	Output CSV file	required
-j	Job: 1 = Prediction, 2 = Motif Search	1

---

Job 1 — CPP Prediction

python CPPLocPred.py -i input.fasta -o results.csv
python CPPLocPred.py -i input.fasta -o results.csv -j 1 -t 0.5 -m /path/to/models/

Flag	Description	Default
-t	CPP probability threshold	0.44
-m	Directory containing model .pkl files	./

Output file columns:

Column	Description
ID	Sequence identifier
Sequence	Amino acid sequence
CPP_Probability	Stage 1 probability (ExtraTrees)
CPP_Prediction	CPP or Non-CPP
Cytoplasm_Probability	Stage 2 localization score
Nucleus_Probability
Mitochondria_Probability
Endo_lysosome_Probability
Others_Probability
Final_Localization	Predicted location(s), semicolon-separated

---

Job 2 — Motif Search

Produces two output files:

• output.csv — one row per sequence (summary)
• output_hits.csv — one row per motif hit (detail)

python CPPLocPred.py -i input.fasta -o motifs.csv -j 2 -l Nucleus
python CPPLocPred.py -i input.fasta -o motifs.csv -j 2 -l Mitochondria -c Koolman
python CPPLocPred.py -i input.fasta -o motifs.csv -j 2 -l Nucleus -c Rasmol --motif_dir /path/to/motifs

Flag	Description	Default
-l	Location: Cytoplasm, Nucleus, Mitochondria, Endo_lysosome, Others	Cytoplasm
-c	Motif class subfolder: None, Koolman, Betts-Russell, Rasmol	None
--motif_dir	Directory containing motif class subfolders	./motifs
--perl	Path to perl executable	perl

Motif file resolved as: <motif_dir>/<class>/<location>_motif

Summary output (output.csv):

Column	Description
ID	Sequence identifier
Sequence	Full amino acid sequence
Length	Sequence length
Total_Hits	Number of motif hits found
Motif_Patterns	Distinct motif patterns matched (; separated)
Hit_Positions	Start-end of each hit (; separated)
Matched_Residues	Matched amino acid strings (; separated)

Detail output (output_hits.csv):

Column	Description
ID	Sequence identifier
Sequence	Full amino acid sequence
Start	Hit start position (1-based)
End	Hit end position
Hit_Length	Length of matched region
Motif_Pattern	MERCI motif pattern
Matched_Residues	Matched amino acid string

---

Input Format

Standard FASTA format:

>seq1
NALAALAKKRQIKIW
>seq2
RQIKIWFQNRRMKWKK
>seq3
GRKKRRQRRRPPQ

Only standard 20 amino acid single-letter codes (ACDEFGHIKLMNPQRSTVWY) are accepted. Sequences with non-standard characters are flagged as Invalid.

---

Examples

# Predict CPPs with default settings
python CPPLocPred.py -i peptides.fasta -o predictions.csv

# Predict with stricter threshold
python CPPLocPred.py -i peptides.fasta -o predictions.csv -j 1 -t 0.6

# Scan for Nucleus-targeting motifs
python CPPLocPred.py -i proteins.fasta -o nucleus_motifs.csv -j 2 -l Nucleus

# Scan for Mitochondria motifs using Koolman class
python CPPLocPred.py -i proteins.fasta -o mito_motifs.csv -j 2 -l Mitochondria -c Koolman

# Scan with custom motif directory
python CPPLocPred.py -i proteins.fasta -o mito_motifs.csv -j 2 -l Mitochondria -c Rasmol --motif_dir /data/motifs

---

Citation

If you use CPPLocPred, please cite:

Raghava et al. (2025) CPPLocPred: Machine learning-based prediction and subcellular localization of cell-penetrating peptides. IIIT Delhi.

---

Web Server

https://webs.iiitd.edu.in/raghava/cpplocpred/

Contact

Raghava Group, IIIT Delhi — https://webs.iiitd.edu.in/raghava/