Generate Model

From DISI
Jump to: navigation, search

This is intended for Shoichet lab setup. But we hope other find it useful. See SEA

How to generate a model to run SEA

We need the files: reference.smi ; reference.set ; reference.fp

First we need to generate the background Z-score distribution:

[prompt]$ sea-background reference.set.gz

This generated the file model.out

Then we need to fit data the data from this background model and build a fit parameter file:

[prompt]$ sea-fit model.out

This generates the file model.out.distfit, containing the values for the best fit of the background distribution to a normal and an extreme value distribution. From this distribution, we need to select the optimal cutoff (in general we look for the similarity value that better fits to EVD) and then use it to get the fit file.

[prompt]$ sea-fit -c cutoff model.out