System setup¶
Requirements¶
The secomo
package is compatible and was tested
with py2.7, py3.4, py3.5 and py3.5.
Prerequisites:
theano
numpy
scipy
joblib
Biopython
matplotlib
pandas
weblogolib
scikit-learn
seaborn
The latter collection of packages are necessary to investigate the model
results using our set of provided functions (e.g. secomo.utils.scatterTSNE()
).
Finally, if possible, we recommend utilizing CUDA. Theano take advantage of cuda, which significantly speeds up the training phase. See Theano documention for more information.