pubmlstget does not work with docker

[thanhleviet]

When I run this command docker run --rm -it -v $(pwd):/data/ sangerpathogens/ariba ariba pubmlstget "Staphylococcus aureus" get_mlst
The get_mlst was expected to be in the current directory I run the command but it was not, event when I replaced $(pwd) with an absolute path, i.e. /home/ubuntu/mlst:/data but It did not work either.

BUT....
If I logged in docker container with docker run --rm -it -v $(pwd):/data/ sangerpathogens/ariba and run ariba pubmlstget "Staphylococcus aureus" get_mlst , it worked, and get_mlst appeared on the host

[martinghunt]

You need to do /data/get_mlst at the end, ie:

docker run --rm -it -v $(pwd):/data/ sangerpathogens/ariba ariba pubmlstget "Staphylococcus aureus" /data/get_mlst

[thanhleviet]

Thank @martinghunt. It works now.
Did I ignore this instruction written somewhere in the wiki?

[martinghunt]

No you didn't, I should probably add example calls when using docker to the wiki...

[thanhleviet]

@martinghunt Though your explanation made it worked but when I executed ariba run get_mlst/ref_db reads_1.fq reads_2.fq ariba_out with the fastq files correctly put in the $(pwd). I have found out that the docker image was not built in a proper way (my sense :)), and it's not quite safe to have root account owning the result folders. So I come up with my Dockerfile which some lines I copied some code from biocontainers

FROM sangerpathogens/ariba

RUN mkdir /data

#Code copied from biocontainers
RUN groupadd fuse &&
useradd --create-home --shell /bin/bash --user-group --uid 1000 --groups sudo,fuse biodocker &&
echo echo "biodocker\nbiodocker\n" | passwd biodocker &&
chown biodocker:biodocker /data

USER biodocker

VOLUME /data

CMD ["/bin/bash"]

WORKDIR /data

With this docker image, I can run smoothly the commands written on the wiki, such as:
docker run --rm -it -v $(pwd):/data/ thanhleviet/ariba ariba pubmlstget "Staphylococcus aureus" get_mlst

I will have a get_mlst folder inside the $(pwd)

docker run --rm -it -v $(pwd):/data/ thanhleviet/ariba ariba run get_mlst/ref_db reads_1.fq reads_2.fq ariba_out

And result folders have non-root permission.

[martinghunt]

You could run this using the existing docker container:

docker run --rm -it -v $PWD:/data sangerpathogens/ariba ariba run /data/get_mlst/ref_db /data/reads_1.fq /data/reads_2.fq /data/ariba_out

ie /data is needed in front of the paths. The resulting ariba_out directory is not owned by root.

[thanhleviet]

Understood! Thank you.
I ran the command:
docker run --rm -it -v $(pwd):/data sangerpathogens/ariba ariba run /data/mlst_get/ref_db /data/data/50_S14_L001_1P.fastq /data/data/50_S14_L001_2P.fastq /data/test1

It works but ....the folder is owned by root
drwxr-xr-x 2 root root 4.0K May 30 13:43 test1

[martinghunt]

That's interesting. When I run it (using a mac) I get this:
drwxr-xr-x 9 mhunt EBI\Domain Users 306B 30 May 13:55 ariba_out
Sorry, but I've no idea why the difference.

[thanhleviet]

Those commands I ran on Ubuntu.
I have just replicated on my mac and the folders are not owned by root.
0 drwxr-xr-x 5 thanhlv staff 170B May 30 14:58 get_mlst/

I guess probably due to Mac uses virtual machine to run docker while Ubuntu uses docker directly. :)
Interesting! Anw, thanks for your help.

[martinghunt]

Thanks. I'll look into changing the docker so it works more easily on ubuntu as well. Thanks for the dockerfile fix!