Random notes: physics

Post date: Feb 9, 2012 1:03:38 PM

Useful quantities

1kBT @ 295 K = 0.586 kcal/mol = 2.452 kJ/mol
1 kcal = 4.184 kJ; 1 kJ = 0.2390 kcal
1 eV = 1.602 e-19 J = 96.49 kJ/mol = 23.06 kcal/mol
1 L = 10²⁷ Å³
Volume of 1 mol of H2O = 18/10³ L = 0.018 L; 1 L of H2O contains 10³/18 mol = 55.5 mol
Volume of 1 molecule of H2O = 2.99e-26 L = 29.9 Å³
[X] (M) = no. of molecules of X / (no. of molecules of H2O/55.6) = 1661 Å³ * no. of molecules of X / V
Additional volume for restraint: d'_i = d_i + sqrt(2*π*k_B*T/k_i)
55.5 is actually 1000/mass(H2O)

Diffusion coefficients

Methods from http://www.ks.uiuc.edu/Training/SumSchool/materials/lectures/6-3-Statistical-Mechanics-Proteins/1-ik_diffusion.pdf

Mean squared path increases linearly with time
1. Velocity autocorrelation decays with tau; D=<v²> tau
2. Velocity distribution of components is a gaussian with 2D stdev
Maxwell distribution

Lifson Roig

Kelley says that the Lifson-Roig helix-coil theory

phi = -60 +- x
psi = -47 +- x
x = 40 deg
helical segment =def= 3 or more consecutive helical residues
each segment has length n-2
helical content =def= sum over all helical segments of (length of segment) / max possible helical length

CGenFF

Upload the molecule to https://www.paramchem.org/index.html .
Save the result as toppar_xxx.str and split to get

top_xxx.rtf
par_xxx.inp

When using autopsf or psfgen, add the top_xxx.rtf file to the list of topology files which are searched.
Combine the parameter file, use in simulation.

Papers

D. Lucent, C. D. Snow, C. E. Aitken, and V. S. Pande, “Non-Bulk-Like Solvent Behavior in the Ribosome Exit Tunnel,” PLoS Comput Biol, vol. 6, no. 10, p. e1000963, Oct. 2010.

Explains how to compute:

water PMF
rotational entropy
local dielectric flucuation tensor
local diffusion coefficient

T. Jain and B. Jayaram, “An all atom energy based computational protocol for predicting binding affinities of protein-ligand complexes,” FEBS Letters, vol. 579, no. 29, pp. 6659-6666, Dicembre. 2005.

Compute binding affinity at +-1 kcal/mol from all atoms.

Convergence of Markov models

Time scales derived from Markov models are affected by the chosen lag time in two ways:

Time scales are reliable only when lag >> λ where λ is the time scale which guarantees memoryless-ness. Until that regime, timescales (typically) increase with lag*.
One can not resolve time scales smaller than the chosen lag. This is due to the fact that all of the motions which occur faster than lag are averaged out.

The above rules apply both to implied timescales (ITS) and mean first passage times (MFPT). The regime for quantitative deductions of timescales (TS) is therefore λ << lag << TS.

* Tentative explanation: one observes what's moving after a lag time τ, 2 τ, etc. If the flux across states does not double when doubling the lag time, it means that the state is still "sticky".

Table of protein temini in CHARMM

NTER has variants GLYP and PROP. ACE has variants ACED, ACP, ACPD.

Random notes: computing

SVN initialization

On one machine only: create a directory (e.g. dir_name) and put initial files in it. Then go to the parent directory and...

svn import dir_name svn+ssh://yourname@tuo/home/svn/lab/papers/dir_name mv dir_name dir_name_old svn co svn+ssh://yourname@tuo/home/svn/lab/papers/dir_name dir_name

On all remaining machines: use the last line only.

Using plumed from VMD

Allow invokation of PLUMED from VMD.

source ~/shared/vmd-utils/vmd_plumed/plumed_proc.tcl set metric [ Plumed::plumed 1 { lig-> [resname LIG] lig<- POSITION LIST <lig> DIR Z } ]

Then, e.g.

multiplot -y $metric -plot

The first argument (1) is the number of CVs. Limitations: does not support PBC. Make sure you have a psf or prmtop loaded if using masses/charges.

Batch plumed on commandline

ls *.dcd| xargs -i+ bash -c '(driver -dcd + -pdb filtered.pdb -plumed META_INP -ncv 3; mv COLVAR +.cv)' n=rama; awk '!/FIELDS/{print FILENAME $0}' *.$n > $n.all rm *.$n sort -k5,5 -g rama.all |less

Acemd restraints

$file --readpdb--> $pdb --loadsystem--+--> $id1 ---+--addgroup---> $gcom1 | | +--> $coor1 +--get_groups---> $glist1 $glist1 ---+---> center_of_mass --> { x y z } | $coor1 ----+

ggplot legends

p+geom_point(aes(1:10,5:14,colour="Pippo"))+ geom_line(aes(4:10,4:10,colour="Pluto"))+ scale_colour_manual("Source",c("blue","red"))

ggplot multiplot

http://gettinggeneticsdone.blogspot.com/2010/03/arrange-multiple-ggplot2-plots-in-same.html

Install python packages as user

Do this stuff in random order

export PYTHONPATH=~/python/lib/python2.5/site-packages/ mkdir -p ~/python//lib/python2.5/site-packages sh setuptools-0.6c11-py2.5.egg --prefix=~/python/ ~/python/bin/easy_install --prefix=~/python/

Install msmbuilder as user

install epd, add epd to path

module add msmbuilder/svn cp -a /shared/lab/milestoning/msmbuilder2/msmbuilder $HOME src=$HOME/msmbuilder dest=/shared/lab/software/msmbuilder_linux-2.0-SVN cd $src LDFLAGS=-L/shared/lab/software/epd-7.1-2-rh5-x86_64/lib/ python setup.py build LDFLAGS=-L/shared/lab/software/epd-7.1-2-rh5-x86_64/lib/ python setup.py install --home=$dest # msmbuilder/msmbuilder/Scripts -> bin cp -a $src/Extras $dest chmod a+x $dest/Extras/InstallExtras.sh cp -a $src/Docs $dest cp -a $src/Tutorial $dest

Generate compressed EPS

convert wallpaper_test.png eps3:wal.eps

Works with input files of any format. The output size should be more or less the same as the input. Add -compress jpeg or -compress zip or -quality XX to force compression if necessary. See http://daicas.net/compressed-images-in-PS-PDF/article.html and http://www.imagemagick.org/script/command-line-options.php?ImageMagick=tbv38oqplsrps97q3p3ga5uus5#compress for details.

CUDA environment variables

http://www.ks.uiuc.edu/Research/vmd/current/ug/node205.html: VMDNOCUDA and

http://developer.download.nvidia.com/compute/cuda/3_1/toolkit/docs/cudatoolkit_release_notes_linux.txt CUDA_VISIBLE_DEVICES

Wrap

pbc set [pbc get -all] -all pbc wrap -center bb -centersel protein -compound res -all

Making videos

Making portable videos is almost impossible. -qscale is because it seems to ignore -b (bitrate). The quality/size ends up being the same for all codecs (for what I'm concerned).

in=sh2p ffmpeg -y -i $in.%05d.ppm -vcodec mpeg2video -qscale 10 $in.mpg ffmpeg -y -i $in.%05d.ppm -vcodec msmpeg4v2 -qscale 10 $in.avi ffmpeg -y -i $in.%05d.ppm -vcodec libx264 -threads 0 -vpre ipod640 $in.mp4