Question 1

What is AMPidentifier?

Accepted Answer

AMPidentifier is a free web tool and Python package that predicts whether a peptide sequence is an antimicrobial peptide (AMP) using an ensemble of gradient boosting, XGBoost and LightGBM models.

Question 2

How do I predict antimicrobial peptides online?

Accepted Answer

Paste one or more sequences in FASTA format at https://www.ampidentifier.com/beta and submit. The tool returns an AMP probability score for each sequence in seconds, with no login required.

Question 3

How accurate is AMPidentifier?

Accepted Answer

On the held-out test set the soft-voting ensemble reaches AUC 0.950, sensitivity 0.949 and specificity 0.784.

Question 4

Can I run AMPidentifier locally or in batch?

Accepted Answer

Yes. Install it with pip install ampidentifier and use the command line interface, or clone the GitHub repository. The web version also accepts CSV export of results.

Question 5

Which types of antimicrobial peptides does it cover?

Accepted Answer

The model is trained on experimentally validated AMPs from public databases, covering antibacterial, antifungal, antiviral and other host defense peptides, against non-AMP sequences.

Question 6

How does AMPidentifier differ from other AMP prediction tools?

Accepted Answer

It combines gradient boosting, XGBoost and LightGBM in a soft-voting ensemble, runs as a web server, a command line tool and a Python package, and exposes the full training code and datasets on GitHub.

Question 7

Are submitted sequences stored?

Accepted Answer

No. Sequences are processed in memory and are not stored on the server.