analysing blat is failing

[Chris Cornelisse]

i made an assembly of arabidopsis using trinity, and trying to run LACE. however, i am getting errors. i figured out where it goes wrong, but not why.

in the function BuildGraph, between #Copy graph before simplifying and ####### Whirl Elimination ######################, i get a runtime error: dictionary changed size during iteration.

and in the first blat run, i sometimes get the error:

add_edge() takes 3 positional arguments but 4 were given

I am working on python 3.6 and used your method to install LACE. any tips?

[Chris Cornelisse]

checked it with the example, but the same happends

[Quarkins]

Hmm interesting, it seems somewhere during runtime a dictionary/list entry is being removed and then trying to be accessed later on. So if you run the test example, you get the same problem: python Lace.py Example/Example_genome.fasta Example/clusters.txt -t -o Test Can you run that command and print exactly what you see, since no one else has had a problem yet (to my knowledge) running the example, i wonder if it is a versions issue (with BLAT or a python package e.t.c). Did you install python and the required packages with Anaconda? Did you install Lace by: cd Lace conda env create environment.yml source activate lace Or by the "less easy mode" Cheers, Anthony

…

On 26 October 2017 at 13:37, Chris ***@***.***> wrote: Reopened #34 <#34>. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#34 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AOTZBy4-JioWDLv4dsnmzrHIth_S6wvWks5swG8EgaJpZM4QHX5h> .

[Quarkins]

It looks like in September NetworkX (a python module for the graph making) has migrated from 1.1 -> 2.0 with some big changes to directed graphs (https://networkx.github.io/documentation/stable/news.html#networkx-2-0) In combination with python going from 3.3->3.6 this changed a fair few things and will mean porting some code to adapt to their changes. For now you can install the Networkx 1.1 version and hopefully that should work!

[Chris Cornelisse]

hi to answer your first question: yes i installed it on the easy mode, not the less easy mode. however, it is installed on python 3.6. i will deinstall module networkx, and reinstall it with version 1.1

[Chris Cornelisse]

the example works now, so will test it on a different dataset. (uninstalled by conda uninstall networkx and installed 1.11 with conda install -c intel networkx)

[Chris Cornelisse]

works, thanks!

[Quarkins]

Fantastic! It seems they changed an awful lot between networkx 1.11 and 2.0 so that it no longer backwards compatible.... Good luck with your supertranscript analysis :)

…

On 27 October 2017 at 11:11, Chris ***@***.***> wrote: Closed #34 <#34>. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#34 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AOTZB9z_O3k9ErVWcsXTF3euMzi1lOptks5swZ5bgaJpZM4QHX5h> .

[Chris Cornelisse]

small edition to this issue:
i changed the environment.yml file to the following:

name: lace
channels:

• bioconda
  dependencies:
• python=3*
• pandas
• numpy
• matplotlib
• blat
• networkx == 1.10

And this works out of the box.
Hope this will help others

[Quarkins]

Thanks for that!
Following our discussion i actually changed the environment.yml to exactly that last week on github. Turns out great minds think alike :)