Combining results from multiple samples for downstream analysis

[eoinoh91]

Hello,

I ran 66 metagenomes through squeezemeta using sequential mode, and on reflection I should have used coassembly as I wish to compare the samples according to the associated phenotypes/metadata.

Regardless, what is the best way to combine them into a single SQM object for downstream analysis in R - is it to go the 'manual' way and first create an individual SQM object in R for each of the 66 samples (after running sqm2zip.py on each output directory), and then combine them using combineSQMlite() in R?

Or is there a more straightforward way to achieve this? Previously I had done something similar using some of the scripts from an older version of SqueezeMeta, but the are now deprecated.

Thanks!

[fpusan]

If you don't need subsetting, then you can just use combine-sqm-tables.py and load the resulting tables with loadSQMlite.
There's a message that combine-sqm-tables.py has been deprecated, since you could achieve the same result with combineSQMlite and I didn't want have to maintain redundant code. But it actually still works fine, and is more convenient to use.

If you want to be able to subset things (and have enough RAM), loadSQM now accepts a vector of project paths as its first argument.

So something like

SQM = loadSQM(c("/path/to/proj1", "/path/to/proj2"))

should combine all the sequential runs in a single object that will for the most part behave as if it came from a coassembly (namely it supports subsetFun and subsetTax)

[eoinoh91]

Thanks for the quick reply!

So I already tried the combine-sqm-tables.py approach - this was my previous method. But I get an error about the KO output files. Here's an example:

`FileNotFoundError: [Errno 2] No such file or directory: 'squeezemeta_annotation_kegg/RMIC-51A/results/tables/RMIC-51A.KO.copyNumber.tsv'

That file is indeed absent in each directory. Has it been renamed in the updated version of SqueezeMeta?

[fpusan]

Those should be there... did you run your project with the --nokegg option?

[fpusan]

Closing due to lack of activity, feel free to reopen

[bheimbu]

Hi @fpusan,

I'm also getting this error, when I use combine-sqm-tables.py (please note, there are really no *KO.copyNumber.tsv file in any project folder I have). I also tried any other method you mention here -- without success

I'm also getting this error for some projects, which is why, I guess, I cannot load all projects via SQM = loadSQM(c("/path/to/proj1", "/path/to/proj2")):

A_dentosus_BEC332 <- loadSQM("A_dentosus_BEC332")
Loading total reads
Loading orfs
    table...
    abundances...
    sequences...
    taxonomy...
    checkm markers...
Loading contigs
    table...
    abundances...
    sequences...
    taxonomy...
    binning info...
Loading bins
    table...
    abundances...
    taxonomy...
Error in names(x) <- value : 
  'names' attribute [7] must be the same length as the vector [1]

Sorry to bother you again.

Cheers Bastian

[fpusan]

No problem at all, that's how bugs get fixed!
Can you share two of the projects, including "A_dentosus_BEC332" and another one?
I'll try to reproduce the error in combine-sqm-tables.py and in loadSQM

[bheimbu]

Of course,

are the zip files sufficient (created with sqm2zip.py) or do you need the original project folders?

Cheers

[fpusan]

The original folders would be better, so I can also test combine-sqm-tables.py

[bheimbu]

Ok, that takes a minute or two. Let you know asap.

[bheimbu]

Here you go.

[fpusan]

Both issues (in combine-sqm-tables.py and loadSQM) have been fixed.
I uploaded a new version of the squeezemeta-dev package (1.7.3.alpha8)