=================
=SuSPect Package=
=================
Christopher M Yates
Centre for Integrative Systems Biology & Bioinformatics
Imperial College London
c.yates11@imperial.ac.uk
www.sbg.bio.ic.ac.uk/suspect

BACKGROUND
----------

SuSPect (disease-SUsceptibility-based nsSNV PrEdiCTor) is a method
for predicting the phenotypic impact of missense mutations.

Using a support vector machine, SuSPect combines sequence and structural
features with systems-level information. In a blind test, we have found SuSPect
to perform significantly better than other methods.

Scores range from 0 (neutral) to 100 (deleterious).

SuSPect is available for use online at www.sbg.bio.ic.ac.uk/suspect
This download is useful if you have large numbers of mutations and would like a
local copy. Scores have been precalculated for the entire human UniProt database
(2013_03 version) and stored in a SQLite database. This database can be queried
directly with SQLite or using the enclosed Perl script (suspect.pl).

The database also includes a table containing scores for nSNPs from dbSNP.

INSTALLATION
------------
Simply untar the downloaded tarball:

tar -zxvf suspect_package-vX.X.tar.gz

This should give a directory, suspect_package, containing:
suspect.pl  Perl script. Run this to get pre-calculated scores
README.txt  This file. If you're reading this, you've probably already done it.
data/	    Various data files needed by suspect.pl, including score database

If you want to use suspect.pl, change the paths at the top of the file:
1) Where you have placed the suspect_package directory,
2) The paths to the directories containing blastall/blastpgp and annovar.

USAGE
-----
suspect.db (in suspect_package/data/) is an SQLite database.
If you wish, you can query this directly using SQLite.
There are two tables, uniprot_201303 and dbSNP.

=uniprot_201303=
Has 23 columns:
uniprot		UniProt accession (6 characters)
pos		Position in the protein
wt		Wild-type amino acid at this position
A-Y		20 columns, one per amino acid, with the SuSPect score for each.

e.g.
Q9NQG7 74 L 38 41 41 34 29 44 41 18 41 12 34 41 44 41 41 31 26 20 41 31

=dbSNP=
Has 5 columns
rsID		The dbSNP rsID for the variant
chr		The chromosomal location of the SNP
uniprot		UniProt accession
nsSNP		Effect of the SNP on the protein
score		SuSPect score for this variant

The same rsID can map to multiple entries in UniProt due to isoforms, etc.
Scores are returned for all entries for which there is data.
e.g.
rs147060589 22:26873014A>C Q9NQG7 L74R 41
rs147060589 22:26873014A>C E5RG08 L69R 51

suspect.pl
----------
suspect.pl can score a file of variants.

suspect.pl takes an input file of nsSNPs and scores them. The input file can be
in one of four formats:
   vcf (see http://www.1000genomes.org/wiki/Analysis/Variant%20Call%20Format/vcf
-variant-call-format-version-4)
   rsID		e.g. rs147060589
   chr		e.g. 22:26873014A>C
   uniprot	e.g. Q9NQG7 L74R #Comments are ignored
   		     E5RG08 69 R #Wild-type amino acid is not needed

To get SuSPect scores, simply run:
   
   perl suspect.pl --input data/test.txt --output test.out 

--input, --in, -i	input file (vcf or UniProt)
--output, --out, -o	output file
--type, -t	 	chr/rsID/vcf/uniprot (default=uniprot)
--verbose, -v
--help, -h		print a help message

REQUIREMENTS
------------
For scoring nsSNPs using UniProt or rsID, there are no further requirements.
For scoring of VCF/chr files:
BLAST (blastall or blastpgp)
ANNOVAR (with hg19 in humandb directory)

CITATION
--------
If you use SuSPect, please cite the following:
   Yates CM, Filippis I, Kelley LA, Sternberg MJE (2014) SuSPect: Enhanced 
   prediction of non-synonymous single nucleotide polymorphism (nsSNP) phenotype
   using network features (Manuscript in preparation)

HELP
----
If you need any further help, or have any comments/queries, email me at:
   c.yates11@imperial.ac.uk