Missing Dependency for Example Script simple_motif_detection.py? about dragonn HOT 10 CLOSED

We'll fix these ASAP. Thanks for reporting.

Anshul

On Aug 10, 2016 11:42 PM, "rmovva" [email protected] wrote:

Hi,

I think there is a discrepancy between this git repo and the zip/tar
files available from http://kundajelab.github.io/. Most noticeably,
the dragonn-0.1.0 zip file contains the example script run.py, which works
fine for me, while the git repo contains the example script
simple_motif_detection.py. Some other files also seem to be different.
2.

When I try to run "python simple_motif_detection.py", I get the
following error message (running on OS X 10.11.6):
Traceback (most recent call last):
File "simple_motif_detection.py", line 5, in
from dragonn.hyperparameter_search import RandomSearch
File "/Library/Frameworks/Python.framework/Versions/2.7/lib/
python2.7/site-packages/dragonn/hyperparameter_search.py", line 6, in
from moe.easy_interface.experiment import Experiment
ImportError: No module named moe.easy_interface.experiment

Is this a missing dependency? I tried installing the relevant module from
http://yelp.github.io/MOE/install.html, but the error persisted.

1. I'm putting this under the same issue as I assume it may be a
   related error, but when I run "dragonn train --pos-sequences
   examples/example_pos_sequences.fa --neg-sequences examples/example_neg_sequences.fa
   --prefix training_example" as indicated under "15 seconds to your first
   DragoNN model", I'm receiving the following error message: Traceback (most
   recent call last): File "/Library/Frameworks/Python.
   framework/Versions/2.7/bin/dragonn", line 9, in
   load_entry_point('dragonn==0.1.0', 'console_scripts', 'dragonn')()
   File "/Library/Frameworks/Python.framework/Versions/2.7/lib/
   python2.7/site-packages/pkg_resources/init.py", line 542, in
   load_entry_point return get_distribution(dist).load_entry_point(group,
   name) File "/Library/Frameworks/Python.framework/Versions/2.7/lib/
   python2.7/site-packages/pkg_resources/init.py", line 2569, in
   load_entry_point raise ImportError("Entry point %r not found" % ((group,
   name),)) ImportError: Entry point ('console_scripts', 'dragonn') not found

Despite these issues, what I have gotten to work so far has been pretty
awesome! This truly is deep learning for dummies. Thanks!
-Rajiv

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#7, or mute the thread
https://github.com/notifications/unsubscribe-auth/AAI7EVoHd5YXbsMWffqb88uVqxE34RL6ks5qesRPgaJpZM4Jh1ig
.

from dragonn.

@rmovva: Thanks for the feedback. Let me address each issue:

The zip/tar files on http://kundajelab.github.io/dragonn/ and the conda package point to the most recent dragonn release, 0.1.0, which can be viewed here. The git repo contains updates, such as the renaming of the script run.py to simple_motif_detection.py, that have not been released yet.

MOE is a missing dependency and its installation doesn't always work as expected. I will try to bundle it with the dragonn conda package, if that doesn't work we will most likely take it out from dragonn.

The dragonn command line interface in "15 seconds to your first DragoNN model" is a recent update that was is not available in the 0.1.0 release in the zip/tar/conda downloads, so you need to update to the version on the git repo. Can you try the following:

conda install dragonn -c kundajelab

which will install the 0.1.0 release. Then clone the git repo and run

python setup.py install

to get updates that haven't been released yet. Does this resolve the issues with the command line interface?

from dragonn.

@jisraeli: thanks for the detailed response. I first deleted both the dragon-0.1.0 release from the zip file and the cloned git repo (which I already had cloned previously) from my machine. I then ran
conda uninstall dragonn
and then installed dragonn again with conda, cloned the git repo again (didn't download the zip file this time), and ran setup.py as you instructed. I then tried
dragonn train --pos-sequences examples/example_pos_sequences.fa --neg-sequences examples/example_neg_sequences.fa --prefix training_example
which still resulted in the error message I noted above.
I then tried
python simple_motif_detection.py
again, which gave me the same error as before with MOE (but that's probably just because you haven't pushed a fix for that yet).

Should I be concerned about the persisting first error message when I run the "15 seconds to your first DragoNN model" command? Seems unlikely, but perhaps I have to do a complete uninstall of all of dragonn's dependencies (since when I ran conda uninstall dragonn it only removed dragonn itself)?

Unrelated to these issues, I had some questions regarding how to use dragonn for other learning tasks than the example case of a single motif (e.g. multi-motif detection, integrating DNase data, etc). What is the best way for us to communicate? (I work nearby, in Beckman, so would be happy to meet if that's easiest).

Thanks for your help!

from dragonn.

@rmovva: That conda install dragonn -c kundajelab followed by python setup.py install after uninstalling doesn't resolve issues with command line interface is not good. I will be working through this issue as we approach our next release next week, we will most likely provide a pypi release to ease installation and updating. Thank you for your feedback, I will keep this issue open until we resolve this.

For questions regarding dragonn usage, it'd be great if you could post your questions on the dragonn-users google group: https://groups.google.com/forum/#!forum/dragonn-users
I'm sure others will ask similar questions, it'd be great if they could find answers on there. Thanks!

from dragonn.

@jisraeli: Thanks! It would be great if you could leave a note here once you've implemented the fixes. A pypi release sounds ideal.

It seems the users group is private, but I sent a membership request. I will post there.

from dragonn.

@rmovva: I will post an update once we work through this.

Just made the group public, we will update the DragoNN site to show the discussions on there.

from dragonn.

@rmovva: in your initial post, "/Library/Frameworks/Python.framework/Versions/2.7/bin/dragonn" doesnt look like a standard installation path for anaconda python. can you check that you are running anaconda python? it is hard for us to reproduce your error without your anaconda environment - could you do a fresh install of anaconda in a user directory and test this again?

from dragonn.

@jisraeli I have the same problem. dragonn-users group shows up as a private group . I sent request to join.

from dragonn.

@ajohi: the dragonn-users group is public so you should be able to post and comment freely. I just checked with one of our lab members and he was able to make a post without sending a request to join. I don't seem to have received requests to join the group this week. can you please try again? I'd be happy to answer your question re pos/neg fastas on there.

from dragonn.

@rmovva: I'm closing the issue for now. Feel free to reopen if you have an update re installation of anaconda.

from dragonn.