Link Search Menu Expand Document (external link)


Is clusTCR available for Windows users?

ClusTCR relies upon the Faiss library, which is officially only supported for OSX and Linux. Windows users may install a WSL in order to properly run ClusTCR.

I wanna use the clustcr-gpu version, but I don’t know if my device has the correct hardware requirements. Is there a way to check this?

clustcr-gpu relies on the Faiss library, which on its turn relies on the cudatoolkit.Only specific GPUs with CUDA-compatibility can be used for ClusTCR’s GPU function. First, you will need to find out which graphics hardware is available in your device. Depending on your OS, you will need to do one of the following things:

> Linux

$ lspci | grep VGA

Alternatively, go to Settings > About > Graphics.

> macOS

For Mac users, go to Apple menu > About this Mac > System Report… > Hardware > Graphics/Displays.

> Windows

Windows users can find graphics card information through Start > Device Manager > Display adapters.

Now that you have identified the type of GPU in your device, you need to check its compatibilities. clustcr-gpu is only supported for NVIDIA GPUs with CUDA capabilities of minimum compute capability 3.5 (source). To check if your GPU is compatible, visit the NVIDIA website.

Multiprocessing gives me RuntimeErrors. How can I solve this issue?

Some users have reported RuntimeErrors when running clusTCR on multiple CPUs. While we are looking into this issue, there is a temporary solution to the problem. Inserting a if __name__ == '__main__': will prevent the creation of subprocesses recursively, thereby avoiding potential RuntimeErrors. Here is a practical example of how you would use this when working with ClusTCR:

from clustcr import datasets, Clustering

if __name__ == '__main__':
     data = datasets.test_cdr3() # Importing some test data
     clustering = Clustering(n_cpus=4) # Multiprocessing with 4 CPUs # Fit data to Clustering object