You are here

make_fragments

3 posts / 0 new
Last post
make_fragments
#1

We are trying to install make_fragments.pl locally still. Does anyone know about the neuralnet file that results from predict_2nd site? It seems to call overrep-3617-IDaa13-7-10-11-10-11-7-7-ebghtl-seeded3-stride-trained.net, but the predict_2nd package seems to have this:
t2k-5651-IDaaHr-5-15-7-15-9-15-13-ebghtl-stride-seeded.net

Does anyone know what effect substituting the neural net file might have?

Thanks,
Christy

Mon, 2009-09-21 12:41
ctaylor

No clue, but this older thread describes running make_fragments in kind of a bare-bones mode. By turning off most of the SS predictors you can get results immediately instead of hammering on all the weird secondary libraries the perl script wants:

http://www.rosettacommons.org/node/1565

Messing with make_fragments.pl is fine for some fragment purposes. If you are doing abinitio, running the full make_fragments.pl might be important because of the intensive SS prediction. If you just need fragments for loop modeling, it doesn't really matter (they're used for sampling and get overwritten as the loop closes). (I don't do abinitio so I'm not sure, but I do loops all the time and I've never cared how good the fragments were, it's all loop anyway!)

Tue, 2009-09-22 06:46
smlewis

> We are trying to install make_fragments.pl locally still. Does anyone know about the neuralnet file that results from predict_2nd site? It seems to call overrep-3617-IDaa13-7-10-11-10-11-7-7-ebghtl-seeded3-stride-trained.net, but the predict_2nd package seems to have this:
> t2k-5651-IDaaHr-5-15-7-15-9-15-13-ebghtl-stride-seeded.net
>
> Does anyone know what effect substituting the neural net file might have?
>
> Thanks,
> Christy

I'm actually trying to do some abinitio work, so the quality does matter. Thanks for the information!

Christy

Tue, 2009-09-22 11:53
ctaylor