BILBOMD

BILBOMD - rigid body modeling using MD conformational sampling

Written by: Michal Hammel & Martin Pelikan

Hardware platforms: in the moment on our LINUX - cluster only

Short user instructions

The program BILBOMD allows determination of three-dimensional domain structure of proteins based on conformational sampling using molecular dynamics (MD) approach.

MD is widely used for exploring conformational space. A common strategy is to perform the simulation at very high temperature (~1500K), the additional kinetic energy prevents the molecule from becoming trapped in a local minimum. Different conformations of the protein are produced at regular intervals along the MD-trajectory. The produced conformations are validate by subsequent calculations of the theoretical SAXS profiles. The program CHARMM is used for MD simulation. The initial atomic models are taken as the starting point of the simulations. In all cases, only the atoms of the linkers rigid domains and loops are allowed to move, while the domains are treated as rigid bodies, with no internal motion. All conformational sampling are performed with the following protocol.

1. Transforming your initial PDB structure to the CHARMM compatible format.

2. The system is subjected to energy Minimization with harmonic constraints on the protein atoms. (MINIMIZED)

3. Heating up to 1500K keeping the protein atoms fixed. (HEAT UP)

4. During the subsequent "phase of production - MD simulation" the resulting structures of the protein were recorded every 0.5ps in a trajectory file. (MD)

5. For each registered conformation the theoretical SAXS profile, the R_G and the corresponding fit to the experimental data were calculated using the program CRYSOL (Svergun et al., 1995). (CRYSOL)

6. Multiconformational model for flexible structures represented by an ensemble containing a number of 2, 3, 4, 5 or 10 different conformations may be explored. The scattering from such an ensemble is readily computed by averaging the individual scattering patterns from the conformers. To select an appropriate ensemble from pool of all (~10000) MD - generated conformations generic algorithm (GA) based search is employed. The conformational divergence of the selected conformations are validate using RMSD and RG. (EOM).

How to start?

In very soon version you will do not need build your initial model, however recent BILBOMD required initial PDB model of your protein.

> ssh sibyls@bl1231.als.lbl.gov # ask MH for acces

> ssh crush

> cd # to your directory where the initial structure (xxx.pdb) and experimental data (xxx.dat) are)

sibyls@crush:eap/test 75% bilbomd ◄┘

*********************************************************
****************** BILBOMD version 1.3 ******************
*********************************************************
****writen by Michal & Martin release 06/2008 ***********
*****should help to everybody who want to used **********
***********MD conformational sampling and EOM ***********
************** rigid body modeling **********************
*********************************************************
*********************************************************
*********************************************************
*********************************************************
*********************************************************
*********************************************************
Do you want transfer pdb to charmm? <N>

Y ◄┘

# 1. BILBOMD need to make own *.psf *crd files. If you respond y he will ask for initial PDBs.
Initial PDBs will be renumbering starting by number 1 for first residue in PDB.
If you have protein complex or oligomer each separate peptide will be separate segment. Please input each segments as individual PDB.

Do you want minimized? <N>
Y ◄┘

# 2. BILBOMD cleans up the sterical clashes and disconnected loops in initial structure.

Do you want heat up? <N>

Y ◄┘

# 3. BILBOMD heat up the minimized structure to 1500K

Do you want MD? <N>

Y ◄┘

# 4. BILBOMD start conformational sampling. You will need to respond few question about how long MD needs to be (one run/ one Rg is eq. to 200 conformations).

Do you want run only CRYSOL? <N> ◄┘

# If you want to validate already existing conformations ( stored in *dcd files) with a another experimental data you can run only this step.

Do you want run EOM (Esemble optimization method)? <N>

Y ◄┘

# 5. BILBOMD can performed ensemble optimization method using saved *fit files. The scattering curves from all the structures in the MD - pool (saved in ../fit/ directory).
He select ensemble with 2, 3, 4 and 5 structure with following output
<filename>ga*_list1.pdb > multiconformational model
avg_optimal_ga*.dat > fit to the experimental data
ga*list.sta > conformational validation molecule/ RMSD/ RG /External_Distance /Internal_Dist.

Numbers of segments ? (segment should be smaller then 1000 residues)
1 ◄┘

# If your initial structure have more then one peptide chain input them as separate segments (separate pdbs).

Enter 1. pdb name ?
eap_c.pdb ◄┘

ATOM 1 N ASN 1 -61.067 21.000 -8.354 1.00 0.00
ATOM 2 CA ASN 1 -60.707 19.703 -7.767 1.00 0.00
ATOM 3 CB ASN 1 -60.384 19.809 -6.267 1.00 0.00
..
..
..

# Your initial structure will be clean up from REMARKS, HETATOM , change MSE to MET,

change dual occupancy residues and renumbered segments (single peptide) starting by no.1.

Enter the output file name ?
eap ◄┘

#Input root filename. Do not used capital letters. Charmm do no like it. In this moment Charmm will build up own
<filename>.pdb , <filename>.crd AND <filename>.psf

Please check the eap for correct Residues ID

# The residues in the initial structure may be renumbered, please check the files for later rigid domains definition.

Do you want make a new const.inp (definition of rigid domains)? <N>
Y ◄┘

#Const.inp file is the file where the rigid domains are defined.
If you already have const.inp from previous BILBOMD run you may editing this file and change start/end residues manually.

Numbers of rigid domains? #Number of rigid domain including fixed domain.
4 ◄┘

Fixed domain - segment ID?
1 #segment for fixed domain
begin ? #Residues range
113
end ?
207
More residues in the fixed domain - segment ID? ◄┘

1. moving domain - segment ID ? #segment for first domain which will be flexible
1 #Residues range
begin ?
1
end ?
98
More residues in 1. moving domain - segment ID ? ◄┘

2. moving domain - segment ID ? #segment for second domain which will be flexible
1 #Residues range
begin ?
222
end ?
316
More residues in 2. moving domain - segment ID ? ◄┘

3. moving domain - segment ID ? #segment for third domain which will be flexible
1 #Residues range
begin ?
327
end ?
421
More residues in 3. moving domain - segments ? ◄┘

Enter the DAT file
experimental.dat ◄┘ # ( ***.dat ) need to be in q , I format

qmax ? <0.3> ◄┘ # Define qmax , default value - 0.3Å-1.

Do you want specify CRYSOL parameter? <N> ◄┘

# You can define dro 0 - 0.08 (contrast of water shell)
and ra (Radius of atom 1.4 - 1.8 Å).
Used default if you want to let CRYSOL to adjust this values.

How many runs (1 - 10 )? # 1 run per 1 Rg is eq. to 200 conformations
10 ◄┘

Rg min? # Define in which Rg range you want screen the conformations
40 ◄┘

Rg max ?
50 ◄┘

Steps between Rg min and Rg max?
2 ◄┘

MD step 0.001-0.002 ? <0.001> ◄┘ # Higher MD step 0.0015 or 0.002 increase the speed of MD simulation but it may be unstable.

BILBOMD should start and perform the operations depending on questions 1,2,3,4,5 .

BILBOMD OUTPUT

crysol.out > validation of all conformations ( filename , Rg, χ²)

../fit/ > directory with all fits

<filename>ga*_list1.pdb > multiconformational model in pdb format

avg_optimal_ga*.dat > multiconformational fit to the experimental data

ga*list.sta > conformational validation molecule/ RMSD/ RG /External_Distance /Internal_Dist.

DCD to PDB

To see your specific conformations (best fit, worse fit etc...) used dcd2pdb command

> dcd2pdb eap 45 1 155000 155000 # Usage: dcd2pdb file RG run Start End

EXAMPLE of segments and rigid bodies definitions for 4 domains , 2 segments protein :