error wen running hsp.py -i COG about picrust2 HOT 10 CLOSED

Hmm how long does this command run for before outputting this error, is it right away? I think there might be highly variable COG gene families that are taking up a massive amount of memory due to many different possible states. I'll look into this and get back to you.

from picrust2.

from picrust2.

Hi Igor,

I believe this problem is related to ~100 COGs which are extremely variable across the genomes. I removed these outlier COGs from the pre-calculated file and this version of the file can be downloaded here (it's on a new branch I'm currently working on).

Let me know if that fixes your problem!

Best,

Gavin

from picrust2.

Hi Gavin,

I'm getting the same error after running the prediction of the KO abundances. I'm running
hsp.py -i KO -t placed_seqs.tre -o KO_predicted -p 4

and I get:
Error in WMPR_ASR_CPP(Ntips = Ntips, Nnodes = tree$Nnode, Nedges = Nedges, :
std::bad_alloc
Calls: lapply ... hsp_max_parsimony -> asr_max_parsimony -> WMPR_ASR_CPP -> .Call
Execution halted
Error running this command:
Rscript /home/camila/Documents/Microbioma_banano/Computacional/picrust2/picrust2/Rscripts/castor_hsp.R placed_seqs.tre /tmp/tmpp7te9isu/subset_tab_9 mp FALSE FALSE /tmp/tmp7ax7z0xz/predicted_counts.txt /tmp/tmp7ax7z0xz/predicted_ci.txt 100

Is there a problem with outliers as well in the KO count table?

Please let me know
Thanks in advance

Camila

from picrust2.

from picrust2.

Thanks Gavin,

I'm analyzing 331 ASVs and I have 32 GB of RAM and also this:
Mem total: 31955
Mem used: 1931
Mem free: 23330
Mem shared: 797
Mem buff/cache: 6694
Mem available: 28773
Swap total: 2047
Swap used: 0
Swap free: 2047

Cheers,

Camila

from picrust2.

from picrust2.

Hi Gavin

Have you had any luck around troubleshooting this issue?
Or do you think it could be due to the header of the first column of the ASV table inputed in the first step? I've always used "#OTU ID" because my txt file comes from a biom file, but before running PICRUSt, I changed the header to "ASV" I don't know if this has anything to do with the error because the 16S and EC run with no error using the same placed_seqs.tre input

Cheers,

Camila

from picrust2.

Hey Camila,

Yes I made some simplifications to the database that should resolve this issue. I will be pushing the new PICRUSt2 version by tomorrow at latest. The name of the first column doesn't matter actually so that's not the problem. In the meantime you could try running hidden state prediction with "pic" rather than "mp" as the hsp_method, which I believe should also resolve this issue.

Best,

Gavin

from picrust2.

Hi again Camila,

I made several updates to the pipeline which should resolve this issue (see PR #39). The key change to fix this error is that I set an upper cut-off of 10 copies of a gene family across all reference genomes. This takes care of the cases where certain reference genomes have 100s of copies of a particular gene family and all intermediate copy numbers 0 ... 100 could have been predicted as separate states. This speeds up the hidden-state prediction step with little or no decrease in accuracy based on my tests.

Please let me know if the new version works for you!

Best,

Gavin

from picrust2.