Quote

A supervised machine learning classification framework to predict potential vaccine candidates (PVCs) specifically against ESKAPE pathogens — trained on biological and physicochemical features with an ensemble Random Forest + Logistic Regression model.

Download VacSol-ML
View Publication
```html
VERSION v1.0
CATEGORY Antigen Prediction
RELEASE DATE Sep 2024
PATHOGENS ESKAPE (6 species)
WEB SERVER Available Free
LICENSE Free
PUBLISHED Vaccine Journal · Sep 2024
```

What VacSol-ML Does

VacSol-ML uses supervised machine learning to classify proteins from ESKAPE pathogens as protective vaccine candidates (PVCs) or non-protective antigens, automating and accelerating reverse vaccinology.

01

Dataset-Driven Training

Built on a curated dataset of experimentally validated protective antigens and non-protective proteins from all six ESKAPE pathogens.

02

Feature Extraction

Extracts biological and physicochemical properties — amino acid composition, dipeptide content, physicochemical indices — from protein sequences.

03

Autoencoder Encoding

Conventional autoencoder-based strategy employed for deep feature encoding and dimensionality-aware selection before model training.

04

Ensemble Model

Final ensemble of Random Forest + Logistic Regression — the two top-performing algorithms — achieves outstanding AUC and WF1 scores.

05

5-Fold Cross Validation

Stratified 5-fold cross-validation applied during training across all seven evaluated ML algorithms to ensure generalizability.

06

Benchmark Validated

Evaluated on a high-quality independent benchmark dataset demonstrating outstanding discrimination between PVCs and non-protective antigens.

07

Web + Standalone

Accessible via web server at vacsolml.mgbio.tech and as a downloadable standalone version for local analysis pipelines.

08

Open Collaborative

Freely available to the research community to encourage collaborative vaccine development against drug-resistant ESKAPE pathogens.

ML Pipeline

Submit a protein sequence and VacSol-ML classifies it through a validated ensemble model.

01

Dataset Construction

A curated dataset of experimentally validated protective antigens and non-protective proteins from all ESKAPE pathogens is assembled as the training corpus
Source: Literature + NCBI

02

Feature Extraction

Biological and physicochemical properties are extracted from each protein sequence — including amino acid composition, dipeptide content, physicochemical indices, and structural features.
Module: FeatureExtractor

03

Autoencoder Encoding & Feature Selection

A conventional autoencoder encodes the high-dimensional feature space and selects the most discriminative features for downstream classification.
Module: AutoencoderEncoder

04

Multi-Algorithm Training (Stratified 5-Fold CV)

Seven ML algorithms are trained: Random Forest, Logistic Regression, SVM, Decision Tree, KNN, Gradient Boosting, Naive Bayes — each under stratified 5-fold cross-validation.
Best: Random Forest · Logistic Regression

05

Ensemble Model & Output

The final ensemble combines Random Forest and Logistic Regression for robust PVC vs. non-protective classification. Results include class label and confidence score.
Output: PVC / Non-PVC + Score

Installation

VacSol-ML is available as a web server and a standalone application for local usage.

Web Server

# No installation required
# Access directly via browser
 
> URL: vacsolml.mgbio.tech
 
 
# Upload FASTA sequence
# Get instant predictions
 
# Supports batch input
✓ Available free · No signup required

Standalone Version

# 1. Download standalone package
# from vacsolml.mgbio.tech
 
# 2. Follow included setup guide
# for your operating system
 
# 3. Run offline on local machine
# — no internet required
 
# Input: FASTA · Output: CSV
✓ Free · Windows & Linux supported

Choose Your Plan

All plans are currently free. Pricing will be announced soon.

FREE

0$ / forever

Full access · No credit card needed

  • ✓ Web server (vacsolml.mgbio.tech)
  • ✓ Standalone tool download
  • ✓ ESKAPE PVC classification
  • ✓ Ensemble ML model (RF + LR)
  • ✓ FASTA protein input
  • ✓ Confidence score output
  • ✓ Single & batch sequences
  •  
Download Free
PREMIUM

100$ / year

Advanced features for labs & institutions

Features coming soon. Stay tuned for updates.

Notify Me

Download VacSol-ML

Free to use. No license required. Choose your preferred access method below

Web Server

Browser-based access, no installation needed

Open Web Server
Standalone Package

Python-based local tool for offline / batch analysis

Download

How to Cite VacSol-ML

If VacSol-ML contributed to your research, please cite the original publication.

VacSol-ML(ESKAPE): Machine learning empowering vaccine antigen prediction for ESKAPE pathogens

PMID: 39126830

DOI: 10.1016/j.vaccine.2024.126204

AMA: Nasir S, Anwer F, Ishaq Z, Saeed MT, Ali A. VacSol-ML(ESKAPE): Machine learning empowering vaccine antigen prediction for ESKAPE pathogens. Vaccine. 2024 Sep 17;42(22):126204.
APA: Nasir, S., Anwer, F., Ishaq, Z., Saeed, M. T., & Ali, A. (2024). VacSol-ML(ESKAPE): Machine learning empowering vaccine antigen prediction for ESKAPE pathogens. Vaccine, 42(22), 126204.
MLA: Nasir, Samavi, et al. “VacSol-ML(ESKAPE): Machine Learning Empowering Vaccine Antigen Prediction for ESKAPE Pathogens.” Vaccine, vol. 42, no. 22, 2024, p. 126204.
NLM: Nasir S, Anwer F, Ishaq Z, Saeed MT, Ali A. VacSol-ML(ESKAPE): Machine learning empowering vaccine antigen prediction for ESKAPE pathogens. Vaccine. 2024 Sep 17;42(22):126204. Epub 2024 Aug 9. PMID: 39126830.

Contact Us

Email Address

Sales: sales@mgbio.tech

General: info@mgbio.tech

Call / WhatsApp

+92 308 0089944
09:00 AM – 05:00 PM

Find Us

NSTP, National University of Science and Technology, Sector H-12, Islamabad

Related Tools by MGBio

AbAMPdb

is the manually-curated database for antimicrobial peptides (AMPs) specifically against Acinetobacter baumannii.The database offers insights of antimicrobial potentials of these AMPs

Read More

Bacterial Immunogens Database

is the manually curated database, where you can find experimentally validated bacterial protective immunogens published and reported in literature. 

Read More

VacSol

a high throughput in silico pipeline to predict potential therapeutic targets in prokaryotic pathogens using subtractive reverse vaccinology

Read More

VacSol-ML

is a pipeline that uses machine learning algorithms to predict potential vaccine candidates in ESKAPE pathogens

Read More

Bvac-AI

expertly analyzes protein sequence data, providing unparalleled insights for the development of targeted vaccines against a wider spectrum of resistant bacteria.

Read More

PanRV

is a Linux based software package in JAVA language that uses both pangenome and reverse vaccinology approaches

Read More

B-Pan

is the updated version of PanRV, a software package for robust pan-genome analysis. B-Pan integrates functional annotations and algorithms for phylogenetic and evolutionary analysis

Read More