CORAL: problem achieving minimum no. of successful steps

Interactive modelling (MASSHA, SASpy) and global minimization programs (SASREF, BUNCH, CORAL, GLOBSYMM)
Post Reply
Message
Author
aleful
Member
Posts: 6
Joined: 2016.11.18 18:26

CORAL: problem achieving minimum no. of successful steps

#1 Post by aleful » 2016.12.02 15:44

Dear SAXS community,

I have a SAXS data on 122 residue long protein and crystal structure of the same protein, but missing 51 residues at N-terminus. The protein is a dimer.
I was trying to model the missing part with CORAL, but with default parameters it fails in the first iteration of simmulated annealing. It doesn't reach the minimum 100 succesful steps, it never reaches more than 20. I assume it is not a good idea to diminish the required min no. of succesful steps, but which parameters should I modify in a first place to make a succesful attempt?

As an input file I use dimeric xtal structure, but do not impose any symmetry in the search. I am using coral, ATSAS 2.7.1 (r6669).
Last edited by ckerr on 2016.12.02 15:56, edited 2 times in total.
Reason: spelling correction in title

ckerr
Active member
Posts: 86
Joined: 2015.03.25 09:03
Location: EMBL Hamburg

Re: CORAL: problem achieving minimum no. of successful steps

#2 Post by ckerr » 2016.12.02 16:00

aleful wrote:As an input file I use dimeric xtal structure, but do not impose any symmetry in the search.
If you do this, CORAL will consider the dimeric structure as one single chain with one 51 residue tail.

I would recommend one of the following options:

1) Use EOM (if the tail is very flexible / disordered)

2) Split the dimer PDB into two PDB files, one for each monomer. Add these both to CORAL, each with its own 51 residue tail, and fix their positions so they do not move relative to one another.

3) Use a single monomer structure and enforce P2 symmetry.

Alex
Active member
Posts: 692
Joined: 2007.08.09 21:10
Location: Planet Earth

Re: CORAL: problem achieving minimum no. of successful steps

#3 Post by Alex » 2016.12.02 18:15

I have a SAXS data on 122 residue long protein and crystal structure of the same protein, but missing 51 residues at N-terminus. The protein is a dimer.
Plot your data first as dimensionless Kratky plot preferably with references like completely unfolded and folded proteins (can be taken from SASDB or peDB) to
understand the level of flexibility. if it's more towards being rigid, use CORAL; if towards flexible - EOM or similar should be used.

aleful
Member
Posts: 6
Joined: 2016.11.18 18:26

Re: CORAL: problem achieving minimum no. of successful steps

#4 Post by aleful » 2016.12.06 22:54

Thank you for your advices, eventually EOM appeared to be more suitable for my system. Next to the dimer scattering data I have also one on a tetramer and as the further step I wanted to model the N-termini of four chains (with EOM). Anyhow I define the scattering mass I got an error: "Alignment Problems: check the sequence and the domains." As input files I use:
- two pdbs, each containing the same xtallographic dimer but positioned differently in the AU (they do not overlap), both pdbs are defined as files with multiple chains;
- single sequence file of the monomer, the same as used for modelling the dimer
In the runs I tried to preform no symmetry was assumed.

It is quite probable, however, that my tetramer is a dimer of dimers, but with no point group symmetry. Is there a way to define it in EOM?

Alex
Active member
Posts: 692
Joined: 2007.08.09 21:10
Location: Planet Earth

Re: CORAL: problem achieving minimum no. of successful steps

#5 Post by Alex » 2016.12.06 23:15

Anyhow I define the scattering mass I got an error: "Alignment Problems: check the sequence and the domains." As input files I use:
use tool "pdb2seq" to check if you have any sequence mismatch or gaps.. eom is sensitive to that.

try to use only one PDB to test where the problem is.
It is quite probable, however, that my tetramer is a dimer of dimers, but with no point group symmetry. Is there a way to define it in EOM?
to my knowledge - no. One can use a trick, e.g.: given you know interface of dimer of dimers, you split it into two separate dimers and use RANCH
on each individually. Once it's done, you can combine models from two RANCH runs in all possible combinations. NB, it's assumed that modelled parts should
not clash with another dimer, else one would need to run additional clash check on combined models to filter out ones with severe clashes.
Voila - you have an external pool of models, whereas each is a dimer of dimers.


HTH, Alex

Post Reply