You are here

Creating model of 89-90 residues by ab initio methodology-help

2 posts / 0 new
Last post
Creating model of 89-90 residues by ab initio methodology-help
#1

Dear rosetta users,
This is my first time using ab initio modelling in order to obtain a protein's structure (though i have some experience on comparative/homology modelling).
I'm trying to obtain the structure of 2 proteins of 89 (COR15A) and 90 (COR15B) residues respectively. Sadly i can't use the homology-modelling methodology because i don't have a proper template to use it.
Anyway, i have already built 100 structures of one of my protein (COR15B) by using the following command:

../../Downloads/rosetta_2014.20.56838_bundle/main/source/bin/AbinitioRelax.linuxgccrelease -database ../../Downloads/rosetta_2014.20.56838_bundle/main/database -in:file:fasta ./COR15B.fasta -in:file:frag3 ./aat000_03_05.200_v1_3 -in:file:frag9 ./aat000_09_05.200_v1_3 -abinitio:relax -relax:fast -abinitio::increase_cycles 10 -abinitio::rg_reweight 0.5 -abinitio::rsd_wt_helix 0.5 -constant_seed -jran 5 -abinitio::rsd_wt_loop 0.5 -use_filters true -psipred_ss2 ./t000_.psipred_ss2 -kill_hairpins ./t000_.psipred_ss2 -out:file:silent COR15B_silent-5.out -nstruct 10 > COR15B-5.log &

I launched it 10 times changing the jran number in order to get the 100 structures, but my problem is that i'm getting really different structures, so basically i don't know which one i should use.
By using the command:

../../Downloads/rosetta_2014.20.56838_bundle/main/source/bin/cluster.linuxgccrelease -database ../../Downloads/rosetta_2014.20.56838_bundle/main/database/ -cluster:radius 3 -in:file:silent COR15B_silent-3.out -out:output -in:file:fullatom

i got 10 different clusters for the 10 initial structures, so each one of them are totally different from the other ones, right?
As i said you before, this is my first time using rosetta, so probably i'm making several things wrongs.
I was looking into the webpage for an specific workflow to use, but i couldn't find :/
So hopefully someone here could guide me on this task. (and also to how to use rosetta in a proper way )
Have a nice day,
Greetings,
Carlos Navarro Retamal
PhD student.

Post Situation: 
Sun, 2014-05-25 12:12
krlitros87

First off, here's the documentation for AbinitioRelax, if you haven't seen it yet: https://www.rosettacommons.org/docs/latest/abinitio-relax.html

Getting very different structures for protein folding is expected. Getting a 20-30 Ang rmsd spread or more is very typical. What you normally do is run a large number of outputs (typically tens of thousands or more, rather than just 100) then do your clustering, and then look at the score-vs-rmsd funnel to your selected structure. You should see that there are low energy structures around your

You may want to take a looks at http://www.meilerlab.org/index.php/jobs/resources under RosettaTutorials.
Particularly http://www.meilerlab.org/jobs/downloadfile/name/IsThisModelGood.pdf and http://www.meilerlab.org/jobs/downloadfile/name/rosetta3.2_folding_prese...
As well as https://www.youtube.com/watch?feature=player_embedded&v=ekTHQcNm9P0 https://www.youtube.com/watch?feature=player_embedded&v=cqORMxg5XoQ and https://www.youtube.com/watch?feature=player_embedded&v=M8r-F59ovwk

In short, one possibility for not seeing a clear trend in your results is that you just don't have enough output structures - you haven't searched thoroughly enough.

(As a final note, if you have any experimental information that might help restrain the structure, that will help to cut down on the search space, and will allow you to get away with fewer output models. - Though typically still in the thousands, rather than in the hundreds.)

Mon, 2014-05-26 09:29
rmoretti