SIFT

I.	INSTALLATION

	Ensure that the instructions in STANDALONE_INSTALLATION 
	has been already been performed.

	Refer to the paths set in <SIFT_HOME>/config/config_env.txt 
	for the following directories:
		<TMP_DIR> = Location of the temporary directory where the results are stored.
		<BLAST_HOME> = Location of the Blast installation.
		<BLIMPS_DIR> = Location of Blimps directory.

II.	DATABASE FORMAT

	This step is required if you are inputting a protein alignment
  
	SIFT searches a database of protein sequences to find homologous 
	sequences.  You will need to download a database of protein sequences
	and format it properly so that SIFT subroutines can read it.

	A.	Database from EMBL:
		ftp://ftp.ebi.ac.uk/pub/databases/fastafiles/uniprot/	

		After downloading the databases, you should have 3 .gz files:
		uniprotkb_swissprot.gz
		uniprotkb_swissprotsv.gz
		uniprotkb_trembl.gz

		In the directory containing the above files, run the following commands:

		>  zcat uniprotkb_swissprot.gz | awk '{if (/^>/) { print ">
		   " $2 " " substr ($_,2,10000)} else { print $_}}' > swiss.uni
		
		>  gunzip uniprotkb_trembl.gz

		>  cat uniprot_trembl | perl -pe 's/\|/\t/g'  | awk ' {if (/^>/) { print ">" $2 } else { print $_}}' | perl -pe 's/\>tr//' > trembl.uni 

		>  cat swiss.uni trembl.uni > swiss_trembl.uni

		>  $NCBI/formatdb -i swiss_trembl.uni -t 'Uniprot-TrEMBL <version>' -p T

		>  $NCBI/formatdb -i swiss.uni -t 'Uniprot-Swiss <version>' -p T

		******************  OR ***********************

	B.	Database from NCBI 
		ftp to obtain ftp://ftp.ncbi.nih.gov/blast/db/FASTA/nr.gz

		Perform this only if you wish to use NCBI's database.

		In the directory containing nr.gz, run the following commands:
		>  gunzip nr.gz

		>  cat nr | perl -pe 's/\|/\t/g' | awk '{if (/^>/) { print 
		   $1 $2} else{ print;}}' > nr_formatted

		>  $NCBI/formatdb -i nr_formatted -t 'NCBI_NR <version>' -p T

		   Assuming that NCBI's blast package (including psi-blast), as well as formatdb is installed in <BLAST_DIR>. 

	The names will be changed to the proper format for proper parsing.

	If you have your own protein sequence database and SIFT is not properly
	recognizing the names, go to <SIFT_HOME>/config/orig_bin/Alignment.c and modify "fix_names"

	Recompile ALL programs.
	Re-run <SIFT_HOME>/config/setup_env.pl perl script as directed in STANDALONE_INSTALLATION Section I Part D. 

      
 
III.	RUNNING SIFT

	A.  Input: Protein sequence. (SIFT chooses homologues).
		Requires 3 inputs:
		1) Protein sequence in fasta format.
		2) Protein database to search.  These sequences are 
		   assumed to be functional
		3) File of substitutions to be predicted on (optional).  
		   See test/lacI.subst for an example of the format. 
		   This file is optional. Alternatively, you can print 
	           scores for the entire protein sequence. 

		Results will be stored in the tmp/<seq_file>.SIFTprediction.

		COMMANDLINE FOR A LIST OF SUBSTITUTIONS:
		If you are in <SIFT_HOME>'s bin directory, the commandline is: 

		csh ./SIFT_for_submitting_fasta_seq.csh <seq file> <protein_database> <file of substitutions> <median conservation- 2.75 recommended>

		EXAMPLE:
		If you have a list of substitutions, type the following:
		csh ./SIFT_for_submitting_fasta_seq.csh test/lacI.fasta <protein_database> test/lacI.subst 2.75

		Results will appear in lacI.fasta.SIFTprediction and 
		look something like:

		K2S     TOLERATED 0.08 3.47 LOW CONFIDENCE
		P3M     TOLERATED 0.08  3.35 LOW CONFIDENCE
		V15K    INTOLERANT 0.00 2.84

		(Note: the results may appear different depending on the version of
		 Uniprot database that you use)

		According to this output, the SIFT score for K2S is 0.08
		and the median information of the sequences that have an 
		amino acid represented at the position 2 is 3.47.  If this
		number exceeds 3.25 the substitution is annotated as 
		having LOW CONFIDENCE (which means too few sequences were
		represented at that position.)  There are enough sequences for
		confidence in the V15K prediction. 


		COMMANDLINE TO PRINT ALL SIFT SCORES
		bin/SIFT_for_submitting_fasta_seq.csh <seq file> <protein_database> - <median conservation - 2.75 recommended>
		A dash "-" replaces the list of substitutions.

		Results will appear in <TMP_DIR>/lacI.fasta.SIFT prediction.  Each row is a position
		in the sequence (row 1 is amino acid position 1, row 2 is amino acid 2) and
	        the SIFT scores for each amino acid substitution are printed for each row.


	B.	Input: Your own protein alignment

		COMMANDLINE FORMAT FOR A LIST OF SUBSTITIONS:
		If you are in SIFT_HOME, the commandline is:
		env BLIMPS_DIR=<blimps_path> bin/info_on_seqs <protein alignment> <substitution file> <output file>

		where BLIMPS_DIR is the path to the Blimps directory set during installation.

		EXAMPLE:
		Type in:
		env BLIMPS_DIR=<blimps_path> bin/info_on_seqs test/lacI.alignedfasta test/lacI.subst test/lacI.fasta.SIFTprediction

		And the prediction results will appear in 
		test/lacI.fasta.SIFTprediction, read above for description 
		of output.

		COMMANDLINE TO PRINT ALL SIFT SCORES:
		env BLIMPS_DIR=<blimps_path> bin/info_on_seqs <protein alignment> - <output file>

		Example:
		Type in:
		env BLIMPS_DIR=<blimps_path> bin/info_on_seqs test/lacI.alignedfasta - test/lacI.fasta.SIFTprediction

		Scores for each position will appear in the file.  

		Read III.A for description of output.

                COMMANDLINE TO PRINT ALL SIFT SCORES
               	csh bin/SIFT_for_submitting_fasta_seq.csh <gi id> - BEST 

		A dash "-" replaces the substitution file, and BEST is optional.
                Results will appear in <TMP_DIR>/<gi # id>.SIFT prediction.

		Read III.A for description of output.