Hi all,
I'm using CS-Rosetta and trying to input a constraint file for specifying backbone NOE's. Rosetta 3.2 works great for dNN constraints, but for some reason it's not recognizing HA atoms. Here's an example couple of lines from my inputNOE.cst file:
AtomPair H 33 H 50 BOUNDED 1.80 5.25 0.50 NOE ;dist 5.000 1.800
AtomPair H 44 HA 38 BOUNDED 1.80 5.25 0.50 NOE ;dist 5.000 1.800
The first line and all of it's kind with "AtomPair H res# H res# ..." works fine. When I try and run my file with dHAN constraints formatted like in the second line, I get:
core.scoring.constraints: Constraint choice: inputNOE.cst
core.io.constraints: read constraints from inputNOE.cst
core.io.constraints: read constraints section --NO_SECTION---
core.io.constraints: no section header [ xxx ] found, try reading line-based format... DON'T MIX
core.io.constraints: read constraints from inputNOE.cst
Exception :
Atom HA 38 not found
protocols::checkpoint: Deleting checkpoints of Abrelax
It seems to me like there's an atom type file somewhere that it must be using to check this information, but I've looked in /home/rosetta/rosetta_source/core and /home/rosetta/rosetta_database without finding it.
Thanks for your help.
I'm going to hazard a guess that you have a centroid protein structure, which doesn't have HA atoms present. Is it in centroid or fullatom?
The types are in rosetta_database/chemical/residue_type_sets/ centroid or fa_standard, depending on if you have centroids or all atoms. I think CS-rosetta is using centroids (at least for the early stages).
I see a patch protein_centroid_with_HA.txt, which looks like it should allow the creation of centroid residues which also have HA atoms. It's active by default, but that doesn't mean the code knows to add HAs. If centroid-ness turns out to be the problem, let me know what your command lines are and I'll see if I can find where the pose is created to tell it to use centroid-with-HA (no promises that I can do it or that it will work).
You can also just fiddle with the distances and use only backbone hydrogens in your constraints.
I think you're right, that it's using centroids, but I'm not totally sure. Here's the script:
#!/bin/csh
#
# CS-ROSETTA: System for Chemical Shifts based protein structure prediction using ROSETTA
# (C) Shen and Bax 2007-2009, Lab of Chemical Physics, NIDDK, NIH
# Version 1.01(build 2009.1117.15)
#
# runRosetta3.com: performs a ROSETTA3.X structure generation
#
# set outDir = `cat paths.txt | awk '$1~"score"{print $2}'`
# set rand_1 = `( ps ax && netstat -a && date ) | sum | cut -c1-6`
# set rand_2 = `( ps ax && netstat -a && date ) | sum | cut -c1-2`
# set rand = {$rand_1}{$rand_2}
${rosetta3} -database ${rosetta3DB} -in::file::frag3 aat000_03.200_R3 -in::file::frag9 aat000_09.200_R3 -in::file::fasta t000_.fasta -in:fix_disulf disulf.cst -constraints:cst_fa_file TLL4_test_bound.cst -constraints:cst_fa_weight 10.0 -abinitio::use_filters false -nstruct 50 -increase_cycles 10 -rsd_wt_helix 0.5 -rsd_wt_loop 0.5 -rg_reweight 0.5 -abinitio::fastrelax -score::weights score13_env_hb -out::nstruct 50 -user_tag j003
Could I also just get it to run in fullatom mode to resolve this issue? If so, what's the command for this? I am currently limited by available cpus, so I was hoping not to have to use fullatom in the fragment-based search. However, if that's the only way to get it to read HA atoms, so be it.
Thank you for your help.
I can't find any code whatsoever related to the centroid_with_HA patch, nor is there a way to run abinitio in fullatom. I've sent an email to the developer who wrote it to see if he has anything to add.
I would try either A) tweak distances and use H for HA, or B) try doing the abinitio with just the H constraints and then add the HA constraints for relaxation later.