DeepTCR is a python package that has a collection of unsupervised and supervised deep learning methods to parse TCRSeq data. To see an example of how the algorithms can be used on an example dataset, see Tutorial.ipnyb. For complete documentation for all available methods, see 'Supervised_Documentation.txt' and 'Unsupervised_Documentation.txt'. While DeepTCR will run with Tensorflow-CPU versions, for optimal training times, we suggest training these algorithms on GPU's (requiring CUDA, cuDNN, and tensorflow-GPU).
DeepTCR now has the added functionality of being able to analyze paired alpha/beta chain inputs. For detailed instructions on how to upload this type of data, refer to the documentation for loading data into DeepTCR.
For questions or help, email: [email protected]
For full description of algorithm and methods behind DeepTCR, refer to the following manuscript:
Sidhom, J. W., Larman, H. B., Pardoll, D. M., & Baras, A. S. (2018). DeepTCR: a deep learning framework for revealing structural concepts within TCR Repertoire. bioRxiv, 464107.
https://www.biorxiv.org/content/early/2018/11/26/464107
DeepTCR has the following python library dependencies:
- numpy==1.14.5
- pandas==0.23.1
- tensorflow==1.11.0
- scikit-learn==0.19.1
- pickleshare==0.7.4
- matplotlib==2.2.2
- scipy==1.1.0
- biopython==1.69
- seaborn==0.9.0
In order to install DeepTCR:
pip3 install DeepTCR
Or to install latest updated versions from Github repo:
Either download package, unzip, and run setup script:
python3 setup.py install
Or use:
pip3 install git+https://github.com/sidhomj/DeepTCR.git