AMPidentifier, antimicrobial peptide prediction

Checking Online: model loaded, predictions ready Offline: backend unreachable, try again shortly Checking server status (Xms) = current /health round-trip latency

AMPidentifier is a toolkit for antimicrobial peptide prediction using ensemble machine learning.

For PyPI: pip install ampidentifier

For terminal use: CLI version

This is the beta layout: Access the stable version

In testing

This round changes the interface only. Models, thresholds and predictions are the same as the stable version.

The front end was rebuilt on a token-based design system: one type scale, one spacing scale and a single set of colour tokens shared by every component, with the layout on a single column and a concentric radius ladder. Controls and panels became glass surfaces with backdrop-filter, keyboard focus rings and reduced-motion fallbacks, the usage map became inline SVG instead of a tile layer, and the result panel carries its state in the URL.

Coming soon: a new batch of trained models will reach the beta before the stable version, and a prediction mode built on a protein language model (PLLM) goes into testing here.

Benchmark, voting ensemble (RF + SVM + GB + XGB + LGBM)
0.950AUC-ROC
0.742MCC
94.9%Sensitivity
78.4%Specificity
Usage
—Sequences classified
—Unique users
—Prediction runs
22Descriptors