DeepTCR scripts for classification of TCRs
Kidman J et al., 2024, Oncoimmunology, Manuscript under review. Sequencing data available GSE222575. Data will become publicly available once manuscript is published. Links will be updated following publication.
These scripts were adapted from DeepTCR and are generated to suit bulk TCRB data from murine cancer models with response to immune checkpoint blockade. TCR data was generated from BALB/c mice bearing mesothelioma (AB1) or renal cell carcinoma (RENCA). Using a bilateral tumour model, mice treated with immune checkpoint therapy (anti-CTLA-4 + anti-PD-L1) were identified as responders (RS) or non-rsponders (NR). Complete data information on murine model is described Zemek et al., Nat Comms 2020 Outputs generated were used in manuscript (Kidman J, 2024, Oncoimmunology, Manuscript under review) for publication. Refer to publication for full description of methods and results.
Refer to DeepTCR for complete install information. See requirements.txt to see package versions used to analyse this dataset.
model_classifer.py & response_classifer.py must be run first before other scripts. model_classifier.py used to train TCRs predictive of tumour model (AB1 or RENCA) based on data from a selected timepoint. response_classifier.py used to train TCRs predictive of therapy response (RS or NR) based on data from a selected timepoint. Outputs from model_classifer.py and respones_classifer.py are required for downstream analyses.
inf_response.py and inf_model.py are used to test the output from respective classifiers on data from other timepoints. dynamic_signatures.py used to plot the weighted proportion of the predictive TCRs in each animal, across all samples. vis_corr_models.py used to correlate the probablity of each TCR being model and response predictive.
The below specifies which script was responsible for generating each figure in the manuscript (Kidman J et al., 2024 Oncoimmunology, under review). Figure S6A - model_classifier.py and response_classifier.py Figure 4D-G - dynamic_signatures.py Figure S6B - inf_model.py Figure S6D-G - inf_response.py Figure SH-I - vis_corr_models.py