Skip to content

raghavagps/NeuroPIpred

Repository files navigation

NeuroPIpred: Prediction, Design and Scan of Insect Neuropeptides

Overview

NeuroPIpred is a machine learning-based computational platform developed for predicting, designing, and scanning insect neuropeptides.

The platform uses experimentally validated natural and modified insect neuropeptides for building predictive models.

NeuroPIpred can:

  • Predict insect neuropeptides
  • Design mutant neuropeptide analogs
  • Scan proteins for neuropeptide regions
  • Perform similarity search using BLAST

Research Paper

Title: NeuroPIpred: a tool to predict, design and scan insect neuropeptides

Authors:
Piyush Agrawal, Sumit Kumar, Archana Singh, Gajendra P. S. Raghava, and Indrakant K. Singh

Journal: Scientific Reports (2019)

Background

Insect neuropeptides are signaling molecules that regulate:

  • Mating
  • Migration
  • Oviposition
  • Metabolism
  • Growth and development
  • Homeostasis

These peptides are promising targets for insect pest control and insecticide development.


Dataset Information

Two datasets were developed:

NeuroPIpred_DS1

  • 875 unique natural insect neuropeptides
  • Random SwissProt peptides as negative dataset

NeuroPIpred_DS2

  • 2024 modified neuropeptides with C-terminal amidation
  • Modified peptides from SATPDB as negative dataset

Machine Learning Algorithms

The following techniques were used:

  • Support Vector Machine (SVM)
  • Random Forest
  • SMO
  • J48 Decision Tree
  • Naive Bayes

Input Features

Amino Acid Composition

Residue frequency-based features.

Dipeptide Composition

Captures residue composition and local residue order.

Binary Profiles

Encodes residue order and positional information.

Split Composition

Uses N-terminal and C-terminal residues separately.


Best Performing Models

NeuroPIpred_DS1

Dipeptide composition based SVM model:

  • Validation Accuracy: 83.71%
  • MCC: 0.67

NeuroPIpred_DS2

Dipeptide composition based SVM model:

  • Validation Accuracy: 97.93%
  • MCC: 0.96

Important Residues

Natural Neuropeptides

Residues enriched:

  • C
  • D
  • F
  • G
  • N
  • S
  • Y

Modified Neuropeptides

Residues enriched:

  • D
  • E
  • F
  • G
  • M
  • N
  • P
  • R
  • S
  • Y

Motif Analysis

Important motifs identified:

Natural Neuropeptides

  • ECC
  • QCK
  • FDEI
  • EIDR

Modified Neuropeptides

  • GPR
  • SFGL
  • WFGP
  • YSF

Web Server Modules

Predict

Predicts neuropeptides from peptide sequences.

Design

Generates mutant analogs for activity optimization.

Protein Scan

Identifies possible neuropeptide regions in proteins.

BLAST

Finds similar experimentally validated neuropeptides.

Download

Allows downloading benchmark datasets.


Technologies Used

  • SVM_light
  • WEKA
  • Random Forest
  • MERCI
  • PSI-BLAST

Applications

NeuroPIpred can be used for:

  • Insect neuropeptide discovery
  • Pest control research
  • Peptide engineering
  • Bioactive peptide design
  • Insecticide development

Performance Comparison

Method Accuracy MCC
NeuroPID 52.57% 0.16
NeuroPIpred 83.71% 0.67

Availability

Web Server:

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

Docker Image:

raghavagps/gpsrdocker


Contact

Prof. Gajendra P. S. Raghava

Department of Computational Biology
Indraprastha Institute of Information Technology Delhi
New Delhi, India

Email: [email protected]


License

Creative Commons Attribution License


Generated from the uploaded NeuroPIpred research paper.

About

NeuroPIpred: Prediction, Design and Scan of Insect Neuropeptides

Topics

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors