SAAPdap/SAAPpred

Installation

Compiling and installing the code

1. Ensure you have the following packages installed system wide:

• R
• perl LWP and LWP::Protocol::https modules
• glibc-static

You can do this by running ./preinstall.sh

2. Ensure you have an uncompressed mirror of the PDB installed; you can use ftpmirror from http://github.com/AndrewCRMartin/bioscripts for this purpose.

3. Modify config.pm as required - see the example config.pm.* files as examples.

4. Run ./install.pl to compile and install the software

Interface install

1. In addition to the standard code install, ensure you have the following packages installed system wide:

• perl JSON module
• wkhtmltopdf

You can do this by running ./preinstall.sh

2. Run ./install.pl interface to install the interface

Directories and files present in the main distribution folder

Files

• config.pm.xxx Example config files
• install.pl Install script
• README.md This README file
• preinstall.sh System install script
• util.pm Utilities used by the install script

Directories

• data/ Data files
• lib/ Perl and R library files
• packages/ C program packages
• plugins/ The analysis plugins
• pred/ The prediction code
• src/ The analysis programs
• www/ The web interface

---

Programs and scripts

init.sh

Initiation script - sets paths etc. - Must be run from current directory. See also setup.sh

setup.sh

Preferred way of setting up all paths etc. As init.sh but can be called from anywhere

pipeline.pl

Basic SAAPdap pipeline program. You would run this on an individual mutation on an individual PDB file.

Use pipeline.pl -h for help

uniprotPipeline.pl

Run SAAPdap pipeline when you have just a UniProt code - it will find all matching PDB files. Calls pipeline.pl

Use uniprotPipeline.pl -h for help

multiUniprotPipeline.pl

Run SAAPdap pipeline on a set of UniProt mutations in a file. Calls uniprotPipeline.pl

Use multiUniprotPipeline.pl -h for help

json2html.pl

Run on a single .json file to generate an HTML file.

No further help available

multiJson2html.pl

Takes the name of a directory containing a set of JSON files and converts them all to HTML and creates an HTML index file summarizing all the mutations in the directory.

Use multiJson2html.pl -h for help

makePDF.pl

Makes a PDF version of all HTML pages in a specified directory

delVoidCache.pl

dump_fosta.pl

---

Moving the install

rm cache/sprot/sprot.idx*
rm plugins/data/specsim.idx*

Dump the fosta database

./dump_fosta.sh
rm ./plugins/data/specsim.idx*

Clean the cache

rm -rf cache/*

or for a specific PDB file:

rm cache/*/*PDBID*

---

Changes to Make

TODO:

• Finish commenting

• Write PQS plugin

• Add to Interface code information on what the interacting protein/ligand is Also to CATH and Pfam annotations

• If the SwissProt cache is out of date then make use of the web based version automatically while rebuilding the cache in the background.

• Code needs to check native residue in the PDB file being examined and warn if it isn't what is expected. e.g. G6PD 1qki has the Arg459->Leu mutation

• Main pipeline code needs to do 1->3 conversion of amino acid names

• Prediction of interface binding energy: http://www.bioinf.manchester.ac.uk/intcalc/

• SAAP::CheckRes() should check that the residue is an amino acid and that it's not CA-only (e.g. P01583 2ila A Asp 137 Asn)

• Uniprot Code should check that the specified native residue is correct

• Need to check for mutant structures and display in interface

• Plugin to find sites adjacent to active and binding sites. Use SwissProt annotation, CSA and Benoit's ligsite.

• Modify SwissProt indexing so that it is not done automatically in the code. If the index is out of date, or if it returns nothing automatically use the web service.

• Check that code relying on the SwissProt index fails correctly if it can't update the index.

• CorePhilic and SurfacePhobic should probably use the same cutoffs for defining hydrophobicity and these should be moved into a config file.

• ./pipeline.pl -u=O00238 -c 6 S 3mdy gives: {"SAAP-ERROR": "UniProt accession code (O00238) not known in PDBSWS"} In fact residues 174-500 of the UniProt entry map to PDB so this should say: "Residue 6 of UniProt accession code (O00236) not found in PDBSWS"

• ./clashes.pl A680 ILE /acrm/data/pdb/pdb2wl1.ent gives: Error: (mutmodel) Missing atoms. No search {"Clash-ENERGY": "", "Clash-BOOL": "OK"} Residue has multiple occupancies throughout (including backbone). Selection of the best atoms then seems to mess up atom order and finding the atoms

• Hopefully can drop FixIE() from json2html.pl and multiJson2html.pl

• Some plugins (e.g. clashes.pl) don't work properly with blank chain names and just freeze.

• On local PDB file, rather than PDB code, Impact gives a 'BAD' result rather than error.

---

Changes Done

• 23.09.11 Make all plugins accept -v and get the pipeliner to pass it into the plugins

• 23.09.11 Split FOSTA.pm into FOSTA.pm and UNIPROT.pm

• 23.09.11 sprotft should use the REST interface to PDBSWS with common modules

• 23.09.11 sprotft.pl should use the UNIPROT module to get sequence data. This should also be modified to deal with the date checking issues currently only done in sprotft.pl

• 23.09.11 Split stuff from config that shouldn't be in there

• 02.11.11 Pipeline program checks to ensure that the residue is found

• 02.11.11 Pipeline program takes an option that the residue number is as specified in UniProt so it looks up the PDB number from PDBSWS

• 04.11.11 Top level pipeline program should do some wrapping of the JSON into a single JSON record

• 29.11.11 Improve main index page from multiJson2html.pl

• 07.12.11 json2html.pl is giving an error HTML page even if only one of the analyses in the contained PDB file has an error

• 09.12.11 BuriedCharge, Interface and SurfacePhobic don't deal correctly with numeric chain names of the form 1.518 e.g. Run ./uniprotPipeline.pl Q13153 L 514 V >People/frances/batch1/data/Q13153_L_514_V.json (or ./pipeline 1.514 V /acrm/data/pdb/pdb2hy8.ent)

• 09.12.11 Summary page from multiJson2html.pl should generate a colour based on the number of structures with a 'bad' analysis rather than being red for any bad.

• 13.12.11 improve the getresol program to get non-crystallography method information more accurately

• 16.12.11 CorePhillic is doing accessibility on the complex not the single protein! (Also SurfacePhobic)

• 23.01.12 specsim lookup is now a web service

• 24.01.12 Write config::CreateXmasFile()

• 07.03.12 CorePhilic should give relative accessibility in the same way that SurfacePhobic does

• 07.03.12 AVP can return <10 voids, so code should pad top 10 voids with zeros e.g.

     1c26 - 4voids
     1sfc,1soh ---- 3 voids
     3bua,1zbq  ----0 voids
     1c25---4 voids
     1sal --- 7 voids
     1iie -- 6 voids
     1p0t -- 8 Voids
     1z6w -- 9 voids
  

• 26.03.12 When plugins create cache directories they need to set permissions for anyone to write to that directory

• 20.06.12 Fixed Internet Explorer compatibility problem with getElementsByClassName()