axd (c46b2)

The AXD Module of CHARMM

David Glowacki and Emanuele Paci, 2010

AXD provides an efficient method for determining simultaneously the
potential of mean force and rates associated to a slow collective variable
of the system (reaction coordinate).

Presently it has only been implemented and tested for the distance
between two atoms, which is the default. Work is in progress to generalize
the method to a number of reaction coordinates (Yew, Glowacki, Paci,
work in progress).

The reaction coordinate is kept within a perfectly reflecting "box" for
a time interval sufficient to thoroughly sample the associated values of the
reaction coordinate. This is done by reversing the velocity of the particles
involved in the reaction coordinate. After a given number of collisions with
the boundaries the reaction coordinate is allowed to increase or decrease so
that a neighbouring box can be sampled. From the number of collision with the
boundaries the potential of mean force over the whole range of reaction
coordinate values can be reconstructed, as well as the absolute rate of
entering or exiting a specific "box".

The method is related to Ron Elber's milestoning but with some crucial
differences. Its application is simpler and requires a single simulation with
a rather simple input to determine the potential of mean force.


* Syntax | Syntax of the AXD command
* Description | Description of the command
* References | References
* Example | Usage Example
* Comments | The current status


Top
[COMMAND SYNTAX]

AXD RC atom-selection [MIN real] [MAX real]
[NBOUND integer EVENTS integer BOUNDS sequence-of-reals]
[IUNJ integer] PRINT_OPTIONS


Top
[DESCRIPTION OF COMMANDS]

[IUNJ integer]
specifies the file in which the value of the rc will be written. The file
contains three columns: 1) timestep; 2) rc value; 3) the inversion boundary
(if velocity inversion occurred at that timestep)

RC
the reaction coordinate definition with which to use AXD. Reaction coordinates
presently available include: 1) DIS (distance between two atoms; this is
the default if no RC is specified)

MIN (real)
the minimum allowed value of the reaction coordinate. When the rc first
becomes greater than MIN, the dynamics will be subsequently constrained so
that the rc does not become smaller than MIN.

MAX (real)
the maximum allowed value of the reaction coordinate. When the rc first
becomes smaller than MAX, the dynamics will be subsequently constrained
so that the rc does not become larger than MAX.

note 1: specification of both MIN and MAX values will, upon first arrival
of the rc in the boundaries, "lock" the reaction coordinate between MIN and MAX

note2: if MIN and/or MAX are not specified, then AXD will expect input for
NBOUND. if MIN and/or MAX are specified in addition to NBOUND, AXD will give
an error.

NBOUND
must be followed by an integer which tells how many real numbers follow BOUNDS

BOUNDS
is followed by a sequence of real values, in ascending or descending order.
These values are the boundaries within which the dynamics will be
consecutively constrained. The order at which the system oscillates through
the bounded boxes is determined by whether the bounds are specified in
ascending or descending order

EVENTS
is followed by an integer that tells the program how many total inversion
events to count (the total on both box boundaries) before progressing to the
next box

PRINT_OPTIONS
control print frequency in unit IUNJ. If PRINT_OPTIONS is not specified,
the default is to print at the frequency specified in the dynamics section of
the input. PRINT_OPTIONS (only one of which may be specified) include:

PRNALL (print to unit IUNJ at every timestep)
PRANGE (real real) which prints to the unit IUNJ at every timestep
when the reaction coordinate is between some range of real numbers.
Note that the real number immediately following PRANGE must be
less than the second value


Top
REFERENCES:

(1) Glowacki, Paci & Shalashilin, J. Phys. Chem. B 2009, 113, 16603-16611
http://pubs.acs.org/doi/abs/10.1021/jp9074898


Top
* CHARMM c36a1 Testcase test/c36test/axd1.inp
* Author: David Glowacki and Emanuele Paci
* Date : June 24, 2010


stream datadir.def

open unit 1 read card name @0toph19_eef1.inp
read rtf card unit 1
close unit 1
open unit 1 read card name @0param19_eef1.inp
read param card unit 1
close unit 1

READ SEQUENCE CARDS
* 13-mer

13
THR TRP ILE GLN ASN GLY SER THR LYS TRP
TYR GLN ASN

GENERATE PEPT WARN SETUP

! { Do not read coordinates }

IC PARAMETERS
IC SEED 1 N 1 CA 1 C
IC BUILD

MINI SD

OPEN WRITE UNIT 43 UNFORM NAME @9axd1.dcd
OPEN WRITE UNIT 44 FORM NAME @9axd1.axd

AXD IUNJ 44 PRANGE 19 21 SELE ((ATOM PEPT 1 N) .OR. (ATOM PEPT 13 C)) END MAX 28 MIN 19

! Test AXD using VV2 integration
DYNAMICS VV2 START NSTEP 2000000 TIMESTEP 0.001 IUNREA -1 IUNWRI -1 KUNIT -1 IUNCRD 43 -
IPRFRQ 100 NPRINT 100 NSAVC 500 INBFRQ -1 ICHECW 0 IEQFRQ 0

STOP


Top
COMMENTS

There are essentially four modes in which AXD may be run:

1) the reaction coordinate (rc) cannot cross an upper boundary (requires
specification of MAX)
2) the rc cannot cross a lower boundary (requires specification of MIN)
3) the rc is maintained between some maximum and minimum value (requires
specification of both MAX and MIN)
4) the rc is confined within a series of boundaries located along the
reaction coordinate (requires specification of NBOUND)

Presently, the method may be used in conjunction with the LEAP and VV2
integration algorithms, and with Langevin dynamics. It works with SHAKE, but
we recommend the SHAKE is not applied to those atoms involvel in the
definition of the reaction coordinate.