MEMPACK Usage Notes
===================

Program  and documentation  is  Copyright  (C)  2009  David T. Jones and
Timothy Nugent, all rights reserved.

All Trademarks and Registered Names are acknowledged in this document.

THIS SOFTWARE MAY ONLY BE USED FOR NON-COMMERCIAL PURPOSES. PLEASE CONTACT
THE AUTHOR IF YOU REQUIRE A LICENSE FOR COMMERCIAL USE.


Compiling MEMPACK
=================

MEMPACK requires the Boost C++ libraries which can be downloaded from
http://www.boost.org/. It was developed and tested using version 1.37 -
later version may not work. On Redhat/Fedora/Centos systems you should
be able to install via yum using a command like:

yum install boost boost-devel

On Debian/Ubuntu systems try:

apt-get install libboost libboost-devel

If available, you can just install the Boost graph library. You may also need
to edit the BOOST variable in the Makefile. Note that this may not compile
without a compatible, early, gcc.

MEMPACK was developed using gcc-4.3 and g++-4.3. You may have to build this
yourself. (You may have more success compiling
https://ftp.gnu.org/gnu/gcc/gcc-4.6.4/). Don't forget to pass an install
location (/usr/local/gcc-4.3.6) to configure. On a compatible Unix or
Linux system, MEMPACK can be compiled simply with:

make

If you've installed Boost in a non-standard location, you may have to
edit the Makefile. And similarly pass the locaton of gcc-4.3.6 to the make
file.

A copy of SVM Light is included; the executable will be placed in the bin
folder where the run_memsat-svm.pl expects to find it. Full details of
SVM light, including the licence, can be found at:

http://svmlight.joachims.org/

To produce graphical representations of the helical packing arrangement,
the GD, GD::SVG and Image::Magick perl modules must be installed. The GD
C library should be present on most modern Linux distributions. Perl
modules can usually be installed by running the following command as root:


cpan -i GD GD::SVG Image::Magick


Otherwise, they can be found at http://www.cpan.org/.

To run mempack you need to download the mempack datasets available at
http://bioinfadmin.cs.ucl.ac.uk/downloads/mempack/ and then:

  ./tar -zxvf mempack_datasets.tar.gz

To configure MEMPACK, the paths to the NCBI binary directory and a database
for PSI-BLAST searches must be set. The script will try to find the right
directory using 'locate blastpgp. If it can't be found you can set the
paths at the top of the run_mempack.pl script:

## NCBI / Database paths
my $ncbidir = '........'; # where blastpgp and makemat are found
my $dbname  = '........'; # e.g. swissprot.fa, which has been formatdb'ed

You can also pass these values using the following parameters:


./run_mempack.pl -d <databse> -n /usr/local/blast/bin/ fasta.fa


Ubuntu Configuration
====================

The ./run_mempack.pl uses bash instead of Ubuntu's dash shell. You'll
have to change it like this:


sudo ln -sf /bin/bash /bin/sh


Running MEMPACK
===============

To run MEMPACK using fasta files, having already set the database and NCBI paths:


./run_mempack.pl -t 45,66,76,97,118,140 examples/1JB0_L.fa


The program requires the transmembrane topology to be passed using the -t
flag. This can be predicted by programs such as memsat-svm.

To run MEMPACK using PSI-BLAST .mtx files:


./run_mempack.pl -mtx 1 -t 38,63,72,96,109,133,153,172,202,224,253,274,286,309 examples/1GZM_A.mtx


Below is a full list of command line paramaters:

Options:

-a <1|2|3>     Distance between atoms in interacting residue pair. Default 1.
               1 = Less than 8 angstroms between C-beta atoms (C-alpha for glycine).
               2 = Less than the sum of their van der Waals radii plus a threshold of 0.6 angstroms.
               3 = Less than 5.5 angstroms between sidechain or backbone heavy atoms.
-mtx <0|1>     Process PSI-BLAST .mtx files instead of fasta files. Default 0.
-n <directory> NCBI binary directory (location of blastpgp and makemat)
-d <path>      Database for running PSI-BLAST.
-t <topology>  Transmembrane topology helix boundaries in the form: 20,35,50,70,91,108
-j <path>      Output path for all files. Default: output/
-w <path>      Directory that contains mempack-svm. Default ''
-e <0|1>       Erase intermediate files. Default 0.
-f <0|1>       Erase files from previous runs. Default 0.
-g <0|1>       Draw schematic. Default 1.
-r <0|1>       Draw residue-residue contacts. Default 1.
-c <int>       Number of CPU cores to use for PSI-BLAST. Default 1.
-h <0|1>       Show help. Default 0.



Example Results
===============

In this  example MEMPACK  is used to predict the helical packing arrangement of
Bacteriorhodopsin. The input file (in FASTA format) is as follows:

>2BRD_A
MLELLPTAVEGVSQAQITGRPEWIWLALGTALMGLGTLYFLVKGMGVSDPDAKKFYAITTLVPAIAFTMYLSML
LGYGLTMVPFGGEQNPIYWARYADWLFTTPLLLLDLALLVDADQGTILALVGADGIMIGTGLVGALTKVYSYRF
VWWAISTAAMLYILYVLFFGFTSKAESMRPEVASTFKVLRNVTVVLWSAYPVVWLIGSEGAGIVPLNIETLLFM
VLDVSAKVGFGLILLRSRAIFGEAEAPEPSAGDGAAATSD

The transmembrane helix boundaries are as follows:

22,43,57,75,93,110,121,140,145,166,187,209,214,235

Run the program like this, having set NCBI and database paths:


./run_mempack.pl -t 22,43,57,75,93,110,121,140,145,166,187,209,214,235  examples/2BRD_A.fa

********************************************************
*   MEMPACK - Predicting transmembrane helix packing   *
*       arrangements using residue contacts and        *
*              a force-directed algorithm.             *
* Copyright (C) 2009 Timothy Nugent and David T. Jones *
********************************************************

Running PSI-BLAST: examples/2BRD_A.fa
/usr/local/blast/bin/blastpgp -a 1 -j 2 -h 1e-3 -e 1e-3 -b 0 -d /home/tnugent/blast-2.2.15/swissprot/uniprot_sprot -i mempack_tmp.fasta -C mempack_tmp.chk >& mempack_tmp.out

bin/svm_classify -v 0 input/2BRD_A_LIPID_EXP.dat models/LIPID_EXPOSURE_ALL.model output/2BRD_A_LIPID_EXPOSURE.predictions
Written output/2BRD_A_LIPID_EXPOSURE.results

bin/svm_classify -v 0 input/2BRD_A_CONTACT.dat models/CONTACT_ALL_DEF1.model output/2BRD_A_CONTACT_DEF1.predictions
Written output/2BRD_A_CONTACT_DEF1.results

Generating layout...
bin/kk_plot output/2BRD_A_CONTACT_DEF1.results > output/2BRD_A_graph.out

Generating JPG image output/2BRD_A_Kamada-Kawai_1.jpg



A number of useful files are produced:

output/2BRD_A_LIPID_EXPOSURE.results : Lipid exposure prediction results. Columns
correspond to residue position, residue type and raw SVM score. Above zero is a
positive prediction (lipid exposed), zero or below is a negative prediction.

output/2BRD_A_CONTACT_DEF1.results : Residue contact and helix-helix interaction
results. Columns correspond to the interacting residue pair, interaction helix
pair and raw SVM scores (only positive results are shown). If you want to constrain
a prediction, you can modify this file by adding interacting residue and helix pairs
and a score of 1. Re-run mempack and a new layout will be plotted using the constrained
results (if the results files exist, svm-classify won't be run again so this step
will be fast).

output/2BRD_A_Kamada-Kawai_1.jpg : JPG images showing the predicted helical packing
arrangement.

output/2BRD_A_graph.out : Helix positions and rotations. Where multiple helical packing
arrangements are produced, they are scored according to the lowest total residue-residue
contact distance.


FINALLY
=======

If you  need  assistance in  getting  MEMPACK  working, or if you find any
bugs, please contact the author at the following e-mail address:

Timothy Nugent
E-mail: [email protected]

Bioinformatics Unit
Dept. of Computer Science
University College
Gower Street
London
WC1E 6BT