ngcrawford / cloudforest Goto Github PK

Hello

Both of the suggested installation options failed on my cluster. I am getting the following errors:

(phyluce) [pisantyg@biocomp-0-3 CloudForest-master]$ pip install git+git://github.com/ngcrawford/CloudForest.git
Collecting git+git://github.com/ngcrawford/CloudForest.git
Cloning git://github.com/ngcrawford/CloudForest.git to /state/partition1/pip-req-build-2lkgKf
fatal: unable to connect to github.com:
github.com[0: 192.30.253.112]: errno=Connection timed out
github.com[1: 192.30.253.113]: errno=Connection timed out
Command "git clone -q git://github.com/ngcrawford/CloudForest.git /state/partition1/pip-req-build-2lkgKf" failed with error code 128 in None

and

(phyluce) [pisantyg@biocomp-0-3 CloudForest-master]$ pip install -U cloudforest
Collecting cloudforest
Could not find a version that satisfies the requirement cloudforest (from versions: )
No matching distribution found for cloudforest

Python, R, git and pip are properly installed and updated (but without Homebrew).

Any suggestions?

Thank you for your help.

Nick,

Some of the tree comparison tests are failing on linux. I'm pretty sure that this is due to very small differences (decimal level) in the expected and observed trees. Similarly, I get another error (can't remember specifically which, but can track down) that results from the expected AIC computation being off from the observed (we require 4 decimal places of identity - we get 1-2).

Pretty sure this is due to platform differences in rand(), etc., but we need to poke around and make absolutely sure that this is the case.

Can you run on some linux to which you have access and see if you get 4-5 failures? We might not even see same issues on Ubuntu and friends vs. Redhat and friends.

Reported to us by Carl Oliveros. During the boostrapping procedure, you can choose to run boostrapping as single-threaded, MPI-threaded, or using multiprocessing. However the bootstrap sampling is not running correctly under multiprocessing - specifically, the code is incorrectly resampling loci and sites across runs, but only during multiprocessing.

I've got a temporary fix in there now, but it's likely not very robust.

Some how I'm getting an extra character at the beginning of the document.

Should be easy enough with argparse.

Command line Usage: cat practice_alignments/*.phylip | ./prep_aligns.py > oneliners.txt

This needs some serious work.
1.) Use dendropy to get the taxa list.
2.) Setup to read in EMR output.

Cut-n-paste code from prep_aligns_genetrees.py

Add a map step that duplicates the oneliner the number of bootstrap replicates, and then does one bootstrap action on each replicate. This should be faster than the current code.