bydmitry / he-erbb2 Goto Github PK

Download link (~3.3Gb)

he-erbb2-github-data
 ├── tissue-samples
 │    ├── [ sample 1 ]
 │    ├──     ...
 │    └── [ sample n ]
 └── models
      └── Her2
           ├── fold-1.pth
           ├── fold-2.pth
           ├── fold-3.pth
           ├── fold-4.pth
           └── fold-5.pth
    

Attach to the docker container and run:

python inference.py -c configs/inference-config.json

The script generates a .csv file with predicted scores.

• Clone the Repository
• CD to the docker folder cd docker
• Edit "DATA_VOLM" path in init-docker.sh - point to the data you downloaded through the linkd, e.g. /your/path/he-erbb2-github-data/. Other variables can remain default.
• Build an image: ./build-docker.sh
• Run a container: ./run-docker.sh
  • Containers run in an interactive mode
  • control + p, control + q will turn interactive mode into daemon mode, i.e. detach from a container without stopping it
  • Reattach container: docker attach [name]
  • Warning: control + a d – detach from TMUX session when inside a container. This will kill the container!

Scientific Reports PDF
https://www.nature.com/articles/s41598-021-83102-6
DOI: https://doi.org/10.1038/s41598-021-83102-6

Bychkov, D., Linder, N., Tiulpin, A., Kücükel, H., Lundin, M., Nordling, S., Sihto, H., Isola, J., Lehtimäki, T., Kellokumpu‑Lehtinen, PL., von Smitten, K., Joensuu, H., Lundin, J. Deep learning identifies morphological features in breast cancer predictive of cancer ERBB2 status and trastuzumab treatment efficacy. Sci Rep 11, 4037 (2021)

This code is freely available only for research purposes. Commercial use is not allowed by any means.