nyuge/rdkit-build repository overview

⁠Concise docker image with RDKit source build

⁠A motivation

What do you feel when you use RDKit with python in docker? I always feel confused in situations in which it is impossible to achieve both the customizability of environments and the smallness of docker containers.

I want to use python of the specific version in small docker container (not in Anaconda), so this source was developed.

⁠Basic environments

• OS: ubuntu16.04 (as the upstream image)
• Python: 3.5.5
• RDKit: Release_2017_09_3

⁠Usages

This image is supposed to be used as a base image for your docker container. But if you use this image, you type a following command

$ docker run -i -t nyuge/rdkit-build

It is good for you to use this image as an upstream image, as follows:

FROM nyuge/rdkit-build

RUN pip install --no-cache-dir --upgrade \
    chainer \
    scikit-learn \
    tqdm
RUN pip install --no-cache-dir --upgrade \
    chainer-chemistry \
    && mkdir -p $HOME/workspace
WORKDIR $HOME/workspace/

⁠Reference

• RDKit repository⁠: The cheminformatics and machine-learning soft ware
• Chainer Chemistry repository⁠: The library for deep learning in biology and chemistry
• Dockerfile for building RDKit⁠: The docker image with python2.7 and source-built RDKit
• alpine-rdkit⁠: The docker image with Miniconda3 and binary-installed RDKit