callumjcparr/nvidia-cuda11.8-dorado repository overview

⁠Creation of image

Image is built on top of nvidia/cuda:11.8.0-cudnn8-devel-ubuntu22.04.

Obtained source for dorado from:

$ git clone https://github.com/nanoporetech/dorado

$ docker run -it --gpus all -v /home/callum/dorado/:/usr/local/src/dorado nvidia/cuda:11.8.0-cudnn8-devel-ubuntu22.04 bash

$ apt-get update && apt-get install -y --no-install-recommends \
        curl \
        git \
        ca-certificates \
        build-essential \
        nvidia-cuda-toolkit \
        libhdf5-dev \
        libssl-dev \
        libzstd-dev \
        cmake \
        autoconf \
        automake

$ cd /usr/local/src/dorado

$ cmake -S . -B cmake-build
$ cmake --build cmake-build --config Release -j
$ ctest --test-dir cmake-build

⁠example usage

$ docker push callumjcparr/nvidia-cuda11.8-dorado:latest

check basic container --rm flag to delete container once you escape the nvidia-smi command with ctrl+c

$ docker run -it --rm --gpus all callumjcparr/nvidia-cuda11.8-dorado:latest watch -n1 nvidia-smi

$ docker run -it --gpus all -v /home/callum/dorado/:/usr/local/src/dorado nvidia/cuda:11.8.0-cudnn8-devel-ubuntu22.04 bash

Remember where you mount the volume is the upper most directory you can access, so you need to consider where your data is stored on the local file system.

if you escape the container you later exec into using this command

$ docker exec -it <container name/id>

$ cmake-build/bin/dorado download --model
$ cmake-build/bin/dorado basecaller --device "cuda:0" --emit-fastq --verbose --min-qscore 10 dna_r10.4.1_e8.2_400bps_sup\@v4.2.0 input > genome_dorado_pass_sup.fastq

You may change chunk_size to 4000 if you are sequencing shorter reads like RNA-seq libraries for dorado to run similar speed to guppy