search

chrisquince / STRONG

Strain Resolution ON Graphs
MIT License
44 stars 9 forks source link

Stuck at bayespaths step #122

Open yuxiangtan opened 2 years ago

yuxiangtan commented 2 years ago

Dear authors,

I run a new test with 20 simulation samples and stuck at the bayespaths step as in the following figure 08f7dc845aaa848499627d7f883892e

This step had run for almost 48 hours already. (My previous try lasted for 5 days and I killed and rerun)

Attached the run.log and the bin_0.log from the bayespath logs folder. run.log Bin_0.log

I am not sure what I can do now since there are no error message.

Best,

Yuxiang Tan

chrisquince commented 2 years ago

Hi Yuxiang Tan,

The program is actually working fine but because it has found a lot of strains in your bins it is performing a rather slow cross validation step. If you want results quicker then rerun just the bayespaths steps (by killing STRONG and renaming the current bayespaths folder) and rerun with the following added to the config file:

optional_args: '--norun_elbow'

within the 'bayespaths' section.

Thanks, Chris

chrisquince commented 2 years ago

PS You may want to leave this though as it is progressing and the cross-validation is useful for getting the true strain number though.

yuxiangtan commented 2 years ago

Hi Chris,

OK, I will wait then.

I run a 5 sample test before, which took around half day. Looks like the run time is not linearly increased.

Thank you!

Yuxiang

rhysnewell commented 2 years ago

Hi @yuxiangtan ,

I've been trying to get STRONG running as well, and get some baseline benchmarks but have been stuck tryign to get a docker image going. It seems you've been far more successful than me, would you be willing to share an updated version of your docker image with spades and such pre-installed in it?

Also, did this ever finish for you? And how did the results look?

Cheers, Rhys

yuxiangtan commented 2 years ago

Hi @chrisquince It is still running and never finished. If it is related to the ELBOW function, does it mean that it fall into a dead loop?

@rhysnewell I just update the docker, which is v1 now and it is the latest one mentioned in https://github.com/chrisquince/STRONG/issues/113

However, it is still a downstream error reported, which I got no answer about how to fix it.

Best,

Yuxiang

chrisquince commented 2 years ago

Hi Yuxiang,

No it is doing what it is meant to which is running the program multiple times at different strain numbers, on my server that will finish in 8 hours but using 100 cores. On real data the ELBO step may be unnecessary anyhow.

We will try and get to that error, do you have info on how to install the docker?

Thanks, Chris

yuxiangtan commented 2 years ago

Hi, Chris

OK, then I will skip the ELBO step. I ran on a server with around 100 cores as well.

What do you mean by "info on how to install the docker"? I just followed your installation instruction within a docker to avoid software conflicts on dependences. Nothing special.

Best,

Yuxiang

DininduSenanayake commented 2 years ago

@yuxiangtan Any chance you can share the actual Dockerfile which you used to build the image yuxiangtan/strong ? (either tag it on DockerHub OR upload to github).
Thanks

yuxiangtan commented 2 years ago

@yuxiangtan Any chance you can share the actual Dockerfile which you used to build the image yuxiangtan/strong ? (either tag it on DockerHub OR upload to github). Thanks

Actually, I uploaded the docker image on DockerHub already. (https://hub.docker.com/repository/docker/yuxiangtan/strong)

However, I am not too sure where to find the Dockerfile,. Could you guide me how I can do it?

Best,

Yuxiang