meren
commented
5 months ago Hey @JiabaoYuuuuu,
I am glad you managed to solve the download issue. We should change the setup function so it dynamically determines whether the user provided a URL online or a file path on their system for these files.
The memory issue is a weird one. You shouldn't need additional software to install -- the error is due to missing mmseqs2 files that were somehow not generated :( Are you submitting your jobs to the server via slurm? Or are you using it interactively?
JiabaoYuuuuu
For the previous problem, I modified the mmseq.py file and changed it to： if mmseqs_profiles_url is None: mmseqs_profiles_url = 'file:///xxxx/PlasX_mmseqs_profiles.tar.gz'
if coefficients_url is None: coefficients_url = 'file:///xxx/PlasX_coefficients_and_gene_enrichments.txt.gz'
Run after plasx setup \ --de-novo-families 'file:///xxx/PlasX_mmseqs_profiles.tar.gz' \ --coefficients 'file:///xxx/PlasX_coefficients_and_gene_enrichments.txt.gz’
Then I run the next step plasx search_de_novo_families \ -g $PREFIX-gene-calls.txt \ -o $PREFIX-de-novo-families.txt \ --threads $THREADS \ --splits 32 \ --overwrite When, the error message is: FileNotFoundError: The file /tmp/tmpienmre36/mmseqs/clu90.m8 was supposed to be created, but it doesn't exist. This might be because the search using mmseqs2 ran out of system RAM. Consider setting the -S flag to reduce the maximum RAM usage. E.g., if you only have ~8Gb RAM, we recommend setting -S to 32 or higher.
My confusion is, do I need to download additional software such as diamond to generate.m8 files? My server has a lot of memory, it should not be because mmseq2 takes up too much memory.