PalMuc / TransPi

TransPi – a comprehensive TRanscriptome ANalysiS PIpeline for de novo transcriptome assembly
Other
26 stars 14 forks source link

Installation Issue (file yml?) #28

Closed francesconardi closed 2 years ago

francesconardi commented 2 years ago

Dear Author, we are trying - with little success - to install TransPi (conda option) but it seems not to create any environment, not even the referenced TransPi environment. First question is whether your recent update is complete or you are still working on it - suggesting the opportunity to wait a while before trying this new version.

A couple of possible issues while trying to go through the initial lines of your precheck script:

Thanks a lot for this useful piece of software, that we are eager to see running on our computers, Regards.

rivera10 commented 2 years ago

Hello @francesconardi,

TransPi now uses individual conda environments for each process. This is because it was having issues when updating packages and working in a single environment. After you clone TransPi and have the databases, you only need to specify -profile conda and the pipeline will handle the creation of the environments for you. That is why it is not using the transpi_env.yml anymore. However, I still have that part in the precheck code (removing soon).

The main issue I see here is that the precheck is not successfully working for the version check of conda. Thanks for pointing that out. I'll make a quick edit so you can continue with the installation. Let me know if you have further issues or questions.

Best, Ramón

rivera10 commented 2 years ago

I made the change. See https://github.com/PalMuc/TransPi/commit/556c2ec9f57ac3a54c173e082981eb34536cc5af

francesconardi commented 2 years ago

Dear Ramon, thanks for your quick fix. One step further. Once the script gets to the point of building the trinotate database (your function trisql_c()), it actually tries to activate an environment named 'TransPi' that is not there (and there should not be, following your previous note). Hence an error. Can you suggest how this can be circumvented? Thanks indeed Francesco

rivera10 commented 2 years ago

Hello, I just merged the dev branch with the master. In the dev I already fixed this issue (see https://github.com/PalMuc/TransPi/pull/29 ). Anything ket me know Best, Ramón

francesconardi commented 2 years ago

Dear Ramon, I am slowly working my way with the installation, manually completing some steps that fail in automatic. Creation of some conda envs failed, and I could rescue the process by issuing the create environment command manually and then resuming the run.

I am now at the rnaquast env, whose install command is: conda create --mkdir --yes --quiet --prefix /home/u3601/prog/TransPi_folder/condaEnv/env-fa8b0a403402a49160ee9bce39b0a3c5 -c conda-forge bioconda::rnaquast=2.0.1=0

The command fails both when issued in the context of the nextflow pipeline (it trows an error) and ALSO if I issue the command manually (unlilke other envs).

Can you suggest a workaround to install rnaquast env, possibly in the same env structure created by TransPi?

Thanks Francesco Nardi

rivera10 commented 2 years ago

Hello @francesconardi,

Sometimes there are issues with conda packages if you already have them. One thing you can try is to use the conda clean --all to erase all packages and info of conda. Then start again the pipeline. Let me know if that works.

Best, Ramón

francesconardi commented 2 years ago

Hi Ramon,

I have succesfully cloned the rnaquast env from a different source: conda create --mkdir --prefix /...../condaEnv/env-fa8b0a403402a49160ee9bce39b0a3c5 -c bioconda rnaquast=2.0.1=0 and, at present, the main line of TransPi is running smootly.

I'm now after the three external packages SignalP, tmhmm and rnammer, where I get some errors. I see that signalp expects input using a different set of options compared to what TransPi issues. I am wondering if this depends on the package version I downloaded. Before I try to edit the nextflow script accordingly, could you indicate the exact version of the three packages you are using (I'm on centos linux). This may be a simple solution.

Cheers, and thanks for your continuing assistance, Francesco Nardi

rivera10 commented 2 years ago

Hello @francesconardi,

No worries at all. Thank you for pointing these issues so I can come with solutions for future updates.

Good to hear the env was created successfully.

For the signalP analysis I am using version 4.1.

Best, Ramón

francesconardi commented 2 years ago

May I know which versions of rnammer and tmhmm you are using as well?

Thank you Francesco Nardi

rivera10 commented 2 years ago

RNAmmer v1.2 and TMhmm v2.0. What errors do you get?

francesconardi commented 2 years ago

Dear Ramon,

I am working on the integration of Rnammer, SignalIP and tmhmm with the main script, and I get a number of errors. The one I am looking into at present arises from the script (apparently) not finding RnammerTranscriptome.pl (output below).

I actually have two copies of the script: ~/prog/TransPi_folder/scripts/RnammerTranscriptome.pl ~/prog/TransPi_folder/condaEnv/env-b96eed500b8c9d678c4cc5d666983118/bin/RnammerTranscriptome.pl

Where is TransPi looking for the aforementioned script? Where is it expected to be?

Thank you, Francesco Nardi

(output)

Command output:

-- Starting with RNAMMER --

-- Done with RNAMMER --

Command error: .command.sh: line 7: RnammerTranscriptome.pl: command not found

rivera10 commented 2 years ago

Hello,

You need to provide the PATH of these programs in the config file. We do this in the pipeline since each user needs a license for using the tools. Once you have the PATH in the config you should be fine https://github.com/PalMuc/TransPi/blob/146e71c19f8337c7a4cce766e83c413b907798ce/template.nextflow.config#L45-L51