A quickstart running on Mac OS X

[ctb]

Bits and pieces of comparem are mildly incompatible with Mac OS X (Sierra, in my case). See this quickstart for getting it working.

I'll provide patches and pull requests as I can, but thought I'd leave this here for others!

[c1imbe2]

ctb, I see from your Quickstart that you're also running CompareM on macOS. I'm running on macOS High Sierra 10.13.3, and have been able to install comparem and run wf_aai without any errors. Both the genecalling and simliarity search seem to produce viable output (at least in query_genes.faa & hits_sorted.tsv (the .dmnd file is unreadable)). However, aai_summary.tsv only contains information for # genes in each genome; #orthologs and AAI info all contain zeros. This occurs with our without the changes in your Quickstart method. So it seems that aai_calculator.py may be the culprit for me (and seemingly others, too)... have you run into any problems such as this? Does anything in particular seem like it could cause this? Thanks

[ctb]

first question - were you able to run it on the data set I tried it on, the one that's included with sourmash? and did you see the same problem? let me know!

[c1imbe2]

Yes, same problem using the sourmash dataset. Terminal gives the following output, regardless if I change the 'sed' and 'sort' commands or leave them as originally written:
sed: 1: "/var/folders/3j/gh2jtvw ...":
invalid command code f sort: unrecognized option `--parallel=8' Usage: sort [-bcCdfigMmnrsuz] [-kPOS1[,POS2] ... ] [+POS1 [-POS2]] [-S memsize] [-T tmpdir] [-t separator] [-o outfile] [--batch-size size] [--files0-from file] [--heapsort] [--mergesort] [--radixsort] [--qsort] [--mmap] [--human-numeric-sort] [--version-sort] [--random-sort [--random-source file]] [--compress-program program] [file ...]
This leads me to believe that there is an alternate 'similarity_search.py' somewhere that is being called by comparem. In fact, I can alter similiarity_search.py however I like -- even delete it -- and I still get the same exact output. A finder search of "similarity_search" only reveals the file "/opt/local/lib/python2.7/site-packages/comparem/similarity_search.py", except for "/opt/local/lib/python2.7/site-packages/comparem/pycache/similarity_search.cpython-36.pyc". Unsure what this file is, but I can alter it too, without consequence. Where is the similarity_search.py file that is being used?!

[ctb]

hi @c1imbe2 try this:

git clone https://github.com/dparks1134/CompareM.git
cd CompareM/
pip install -e .

and then modify the file. My instructions were wrong - you don't want to do a regular pip install!

[ctb]

(The pip install -e installs it in developer mode, where changes to the source code will be reflected in the execution.)

[c1imbe2]

I uninstalled using pip uninstall comparem and re-installed using pip install -e comparem. Same result. I can delete similarity_search.py and comparem still runs fine.

[ctb]

Sorry - do git checkout, cd into dir, do pip install -e in that dir! :)

…

-- Titus Brown, ctbrown@ucdavis.edu

On Mar 15, 2018, at 1:01 PM, c1imbe2 ***@***.***> wrote: I uninstalled using pip uninstall comparem and re-installed using pip install -e comparem. Same result. I can delete similarity_search.py and comparem still runs fine. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

[c1imbe2]

@ctb Thanks -- this worked for me. After installing in developer mode, the trick was editing the simliarity_search.py found in the the python2.7/comparem directory, not the directory containing cloned files. Thank you!!

[ctb]

great!!

[FWittmers]

I am sorry @ctb but I am not able to get comparm working on my mac.
I tried to follow your guide https://hackmd.io/L2llRUU_SrWfI4OYN-uozQ?view and also this conversation, but no matter what I do I end up with 0 Orthologs (which is the problem leading to all this).
I guess my main problem is that I do not understand how I am supposed to edit the similarity_search.py file. I am biologist and trying to learn bioinformatics on my own for the past few month, but I have no idea what is going on exactly and which sort/gsort I have to change. Would be great If you could clarify how to solve all this.
The KostaLab alternative is not an option for me since I have more than their allowed max genomes.