AbInitio: Predict structure from sequence

KEYWORDS: STRUCTURE_PREDICTION GENERAL

Run on linux

(where $ROSETTA3=path-to-Rosetta/main/source)

$> $ROSETTA3/bin/AbinitioRelax.default.linuxgccrelease @flags

Run on macs

Rosetta/main/source/bin/AbinitioRelax.macosgccrelease @flags

Flags

-in:file:fasta ./input_files/1elwA.fasta
-in:file:frag3 ./input_files/aa1elwA03_05.200_v1_3
-in:file:frag9 ./input_files/aa1elwA09_05.200_v1_3
-in:file:native ./input_files/1elw.pdb

-abinitio:relax
-nstruct 1
-out:pdb

-use_filters true
-psipred_ss2 ./input_files/1elwA.psipred_ss2
-abinitio::increase_cycles 10
-abinitio::rg_reweight 0.5
-abinitio::rsd_wt_helix 0.5
-abinitio::rsd_wt_loop 0.5
-relax::fast

Analyze Output

The output_files directory contains example output.

In score.fsc get a score and RMS for each model. Typical analysis makes a scatter plot of these with RMS on the x-axis and score on the y-axis. Look for a "funnel" to low energies and RMS in a successful ab initio prediction. A failed prediction will not have low RMS/energy structures. For the following arguments for full production run (using the minirosetta compile):

-abinitio::fastrelax
-abinitio::increase_cycles 10
-abinitio::rg_reweight 0.5
-abinitio::rsd_wt_helix 0.5
-abinitio::rsd_wt_loop 0.5
-abinitio::use_filters false
-database minirosetta_database
-ex1
-ex2aro
-frag3 aa0000103_05.200_v1_3
-frag9 aa0000109_05.200_v1_3
-in:file:boinc_wu_zip ploop23_control_fold_data.zip
-in:file:native 00001.pdb
-mute all
-mute all
-out:file:silent default.out
-relax::default_repeats 15
-silent_gz

resultfiles = default.out.gz queue = 3000

And run a relax:

-database
-ex1
-ex2aro
-frag3 aa0000103_05.200_v1_3
-frag9 aa0000109_05.200_v1_3
-in:file:boinc_wu_zip ploop23_control_fold_data.zip 
-in:file:fullatom
-in:file:native 00001.pdb
-in:file:s 00001.pdb
 minirosetta_database
-mute all
-out:file:silent default.out
-relax::default_repeats 15
-run:protocol relax
-silent_gz

resultfiles = default.out.gz