larmord (c46b2)


LARMORD: A Distance-based Chemical Shift Predictor

By Aaron T. Frank, Sean M. Law and Charles L. Brooks III

LARMORD is a method to calculate chemical shifts based on inter-atomic distances
between a nucleus of interest and its neighboring atoms. The main reference
for the LARMORD method is:

(1) Frank, A.T, Law, S.M, and Brooks III, C.L., "A Simple and Fast Approach for Predicting
1H and 13C Chemical Shifts: Toward Chemical Shift-Guided Simulations of RNA",
Manuscript in Preparation.


* Syntax | Syntax of the LARMORD command
* Background | Description of LARMORD methods
* Examples | LARMORD usage examples


Top
[Syntax LARMORD]

!setup LARMORD

LARMORD { larmord-spec } ! setup LARMORD
{ CALCULATE } ! calculate chemical shifts using current coordinates
{ CLEAR } ! clear LARMORD data structures
{ ON/OFF } ! turn LAMORD restraint calculation on/off
{ RESET } ! reset LARMORD option flags and optionally read in new parameters

larmord-spec::= { SCALE real }{ CUNIT fortran unit }{ LUNIT fortran unit }{ HARM | LOG TEMP real }{ shape-spec }{ weight-spec } SELE {atom selection} END
shape-spec::= [FLAT]
weight-spec::= [WT1] [WT2]

Keyword Default Purpose

SCALE 1.0 Force constant for chemical shifts restraint to term

CUNIT false Fortran unit from which to read in chemical shifts data

LUNIT false Fortran unit from which to read in LARMORD parameters

HARM none Use a harmonic (mean-squared-error) restraint

LOG none Use the Bayesian inspired log-harmonic restraint term (see Habeck et. al PNAS 103 (2006): 1756-1761)

TEMP 300.0 Temperature to be used in the pre-factor of the log-harmonic restraint term (see Habeck et. al PNAS 103 (2006): 1756-1761)

FLAT none Use flat-bottom potential. No restraint forces are applied in error is less than the expected
error as specified in the measured chemical shifts file (see below).

WT1 none Use a 1/MAE weighting factor used to account for the accuracy of the predictors for the various nuclei. Default use a constant weight (i.e. weight factor = 1.0).

WT2 none Use a R/MAE weighting factor used to account for the accuracy of the predictors for the various nuclei. Default use a constant weight (i.e. weight factor = 1.0).

Top
Background

LARMORD implements a simple and fast model for predicting chemical shifts. Specifically, the chemical shift, \delta_{i}^{pred}, for a given
nucleus i, is given by a polynomial expansion of inter-atomic distances, i.e.,

\delta_{i}^{pred} = \delta_{i}^{rc}+ \sum_{j}^{N}α_{j} r_{ij}^{beta} Eq.(1).

Here, \delta_{i}^{rc} is the so-called “random coil†chemical shift for nucleus i, N is the total number of
heavy atoms in the RNA, r_{ij} is the inter-atomic distance between atoms i and j, and alpha_{j} and beta_{j} is parameter
that depends on the atom and residue type associated with j (Note: in ref XX beta = -3 for all j(s)). Since the module was
developed initially for RNA, the provided parameters (toppar/larmord.dat) are specifically for predicting
chemical shifts for non-exchangeable 1H (H1’, H2’, H3’, H4’, H5’, H5’’, H2, H5, H6, H8) and
protonated 13C (C1’, C2’, C3’, C4’, C5’, C2, C5, C6, C8) RNA nuclei. That being said the current implementation is general and so
LARMORD can in principle be used to predict chemcial shifts for any biomolecule, provided the parameters are available
(alphas and betas Eq. 1).


The LARMORD parameter file contains the alpha and beta that are used in Eq. 1. For example, let say one wants to
calculate chemical shifts for a C1' the corresponding parameters may look like:
<integer, NLCS> ! number of parameter records contained below
...
C1' GUA C1' 0.428401203243 -3
C1' GUA C2' 0.982419947699 -3
C1' GUA C3' 2.76813085461 -3
C1' ADE C1' 1.51229359219 -3
C1' ADE C2' 1.72348040483 -3
C1' ADE C3' 0.288474060452 -3
C1' ADE N9 -0.0486718470351 -3
C1' URA C1' -0.0 -3
C1' URA C2' -0.189115925066 -3
C1' URA C3' 1.41917076395 -3
C1' CYT C1' 0.0223761113264 -3
C1' CYT C2' 0.468384222507 -3
C1' CYT C3' 2.66350021597 -3
C1' CYT C4' 2.52513704899 -3
...
For a given entry:
Column 1: nucleus that the parameters correspond to (in this example, C1')
Column 2-3: residue name and atom name, respectively of neighboring atom (atom j in Eq. 1)
Column 4-5: alpha and beta corresponding to j (see Eq. 1)

NOTE: This parameter file is stored in toppar/larmord/larmord_rna.dat

In addition to the parameter file, LARMORD also requires the user to input chemical shift list file that specifies which
nuclei chemical shifts should be calculated for. For example:
<integer, NCS> ! number of chemical shift records for restraint comparison
...
ADE 9 C1' 87.754 93.1 0.85 0.53
ADE 20 C1' 93.647 93.1 0.85 0.53
ADE 21 C1' 92.688 93.1 0.85 0.53
ADE 17 C1' 91.571 93.1 0.85 0.53
ADE 26 C1' 92.508 93.1 0.85 0.53
ADE 15 C1' 91.785 93.1 0.85 0.53
ADE 12 C1' 93.037 93.1 0.85 0.53
ADE 9 C2 154.648 154.3 0.73 0.49
...

For a given entry:
Column 1: CHARMM residue name
Column 2: CHARMM residue number
Column 3: CHARMM atom name
Column 4: measured chemical shift
Column 5: "random coil" chemical shifts
Column 6: expected error for that particular kind of nucleus (MAE or RMSE)
Column 7: expected correlation (Pearson correlation)

NOTE: Though mandatory, column 6 and 7 are only important when using WT1 or WT2 keywords (see above and reference XX)


Top
Examples

1) Setup and calculate chemical shifts using LARMORD
...
OPEN READ FORM UNIT 10 NAME "DATA/SHIFTS.DAT"
OPEN READ FORM UNIT 11 NAME "DATA/LARMORD.DAT"

LARMORD CUNIT 10 LUNIT 11 SCALE 1.0 HARM SELE SEGID RNA01 END
LARMORD CALCULATE
ENERGY
SET CSERROR = ?CSRE

NOTE: The chemical shift error (which is the same as the restraint energy when used in the context of MD simulations) is stored in ?csre.


2) Setup and run CS-restrained MD simulations using a weighted log-harmonic restraint:
...
! LARMORD
OPEN READ FORM UNIT 10 NAME "DATA/SHIFTS.DAT"
OPEN READ FORM UNIT 11 NAME "DATA/LARMORD.DAT"
LARMORD CUNIT 10 LUNIT 11 SCALE 1.0 LOG WT2 SELE SEGID RNA01 END


! Dynamics
OPEN UNIT 21 WRITE UNFORM NAME SCRATCH/HEAT.DCD
OPEN UNIT 22 WRITE FORM NAME SCRATCH/HEAT.RES

SHAKE BONH TOL 1E-09 PARAM

DYNAMICS LEAP START -
TIMESTEP 0.002 -
NSTEP 10000 -
NPRINT 500 -
IPRFRQ 10000 -
NSAVC 500 -
ISVFRQ 1000 -
IUNREAD -1 -
IUNCRD 21 -
IUNWRI 22 -
FIRSTT 250 -
FINALT 325 -
TSTRUC 325 -
TBATH 325 -
ICHECW 0 -
IHTFRQ 100 -
IEQFRQ 0 -
IASORS 1 -
IASVEL 1 -
ISCVEL 0 -
INBFRQ -1 -
ILBFRQ 0 -
IMGFRQ -1 -
NTRFRQ 1000 -
ECHECK -1


Finally, see test case c40test/lamord_test1.inp for additional examples.