StriMap: Discovering TCR-HLA-Epitope interactions with Deep Learning

strimap-tools is a package for analysis of peptide-HLA presentation and TCR specificity. It is designed to help researchers understand the interactions between T cell receptors (TCRs) and peptides presented by human leukocyte antigen (HLA) molecules, which play a crucial role in the immune response.

To facilitate use by biologists and help bridge the gap between the machine-learning and immunology communities, we developed an accessible web portal www.strimap.com that enables predictions and finetune models on your own data.

If you prefer to run the package locally, please follow the instructions below to install strimap-tools.

Installation

Create a conda environment. strimap-tools requires Python 3.9.

conda create -n strimap-env python=3.9
conda activate strimap-env

Download the source code from GitHub and install the package along with its dependencies.

git clone https://github.com/uhlerlab/strimap-tools.git
cd strimap-tools
pip install -r requirements.txt

Usage

Training data and pre-trained models

Pre-trained models and training data can be found at: zenodo

Train and predict peptide–HLA presentation with StriMap

A complete, reproducible workflow for training and prediction is provided in:

📓 phla_predictor.ipynb

This notebook demonstrates how to:

• Train a peptide–HLA (pHLA) presentation predictor with 5-fold cross-validation
• Load a trained checkpoint and run prediction/inference (and optional evaluation if labels are available)

Expected input (CSV format):

Column	Description	Example	Note
peptide	Peptide amino acid sequence	GILGFVFTL	Required
HLA	HLA allele	HLA-A*02:01	Required
label	Peptide–HLA presentation label (0 or 1)	1	Required for train/val

Train and predict TCR-pHLA specificity with StriMap

A complete, reproducible workflow for training and prediction of TCR–pHLA specificity is provided in:

📓 tcrphla_predictor.ipynb

This notebook demonstrates how to:

• Train a TCR–pHLA specificity predictor using cross-validation
• Load trained checkpoints and perform prediction/inference on new TCR–pHLA pairs

⚠️ Warning The TCR–pHLA predictor is built upon a pre-trained pHLA presentation model. You must provide a trained pHLA model checkpoint or load one of our pretrained models before training the TCR–pHLA specificity model.

Expected input (CSV format):

Column	Description	Example	Note
cdr3a	Alpha chain CDR3 sequence	CARRGAAGNKLTF	Required
cdr3b	Beta chain CDR3 sequence	CASSPSAGDYEQYF	Required
Va	Alpha variable gene	TRAV24*01	Required
Ja	Alpha joining gene	TRAJ17*01	Required
Vb	Beta variable gene	TRBV4-3*01	Required
Jb	Beta joining gene	TRBJ2-7*01	Required
peptide	Target peptide sequence	LLWNGPMAV	Required
HLA	Target HLA allele	HLA-A*02:01	Required
label	TCR–pHLA binding label (0 or 1)	1	Required for train/val

Citation

If you use strimap-tools in your research, please cite the following paper: