Recipe for generating proteotypic peptides for example database
From SPCTools
Notes creating proteotypic peptide information - based on Human 2009-04 modified to include changes needed to show (almost) all peptides. Updated based on 2009-08 mouse build, see specific build README below for additional info and examples. Root directory for this is at /net/db/projects/PeptideAtlas/species
1. Set up reference database file
# Set up add a couple dirs to your PATH, bash syntax export PATH=/regis/sbeams/bin/:/package/genome/tmhmm_sigp_wrapper/:/net/db/projects/PeptideAtlas/species/bin/:$PATH
# Make processing dir, cd there, and assemble source data. cd /net/db/projects/PeptideAtlas/species mkdir organism mkdir date
# Get database file to work on - see Mouse build instrux below if this needs to be assembled. cd /net/db/projects/PeptideAtlas/species/organim/date cp reference_db.fasta .
# Assuming accessions are correct, filter decoys and trim long proteins longer than 8999 AA (which choke Peptide Sieve) processFasta.pl -f reference_db.fasta -r 'DECOY_' -e -o reference_db_no-decoys.fasta
# Break files into bite-sized chunks! split_fasta.pl --entries 10000 --filename_root input_split reference_db_no-decoys.fasta
2. Run predictor algorithms.
A) Peptide Detectability Predictor
- symlink binaries. ln -s /net/db/src/DetectabilityPredictor/Standalone/PeptideDetectabilityPredictor ln -s /net/db/src/DetectabilityPredictor/Standalone/stand.bin
- run wrapper scripts that automate the searching, located in /bin/net/db/projects/PeptideAtlas/species/bin run_PDP.csh runs predictor on each sub-file - Once the run is complete... mk_PDP.csh merges results files into results.PDP
B) Run Peptide Sieve
- symlink binaries. ln -s /net/db/projects/PeptideAtlas/ExternalData/proteotypic/bin/PepSieve_20080530/peptideSieve_080527/peptideSieve/bin/PeptideSieve . ln -s /net/db/projects/PeptideAtlas/ExternalData/proteotypic/bin/PepSieve_20080530/peptideSieve_080527/peptideSieve/properties.txt .
- run wrapper scripts run_PS.csh runs predictor on each sub-file - Once the run is complete... mk_PS.csh merges results files into results.PS
3. Merge and process results from the two prediction engines.
- Merge predictions /regis/sbeams/bin/mergeProteotypicScores.pl -f reference_db_no-decoys.fasta -p results.PS -d results.PDP -o proteotypic_merged.tsv
- Sort with ENS entries first for mapping, since the same peptides from proteins without mapping can then borrow the mapping info. - (For yeast use the -y flag, other non-ENS organism flags may be needed. ) sortEnsFirst.pl proteotypic_merged.tsv > proteotypic_merged-sorted.tsv
- Calculate genome mappings. If calc script is invoked with no args, it will output a usage stmt that includes - current (2009-09) ENS mapping file options ( e.g. -d Mus_musculus_core_52_37e ) nohup calculateNGenomeMappings.pl -f reference_db_no-decoys.fasta -p proteotypic_merged-sorted.tsv -o proteotypic_merged-sorted-mapped.tsv -d species_core_52_37e &
- End general build notes section
The next step is to load these into the database for later use. This requires that the reference database used is already loaded
as a biosequence set. The load_theo_proteotypic_scores.pl script is located in the SBEAMS/PeptideAtlas codebase in $sbeams/lib/scripts/PeptideAtlas.
nohup load_theo_proteotypic_scores.pl -i merged_results_sorted_mapped_final.tsv -s Hs_ENSP_IPI_SPvarspl_decoy &
Usage statements for various scripts
Usage: load_theo_proteotypic_scores.pl [OPTIONS]
Options:
--verbose n Set verbosity level. default is 0
--quiet Set flag to print nothing at all except errors
--debug n Set debug flag
--testonly If set, rows in the database are not changed or added
--list If set, list the available builds and exit
--help print this usage and exit.
--purge_mappings Delete peptide mappings pertaining to this set
--set_tag Name of the biosequence set tag
--update_peptide_info will update info in proteotypic_peptide table, e.g.
pI, mw, SSRCalc, Peptide Sieve. Does *not* currently
update info in proteotypic_peptide_mapping table, so
one should run purge_mappings first and then update.
--input_file Name of file with PepSeive and Indiana scores, as
well as n_mapping info.
e.g.: load_theo_proteotypic_scores.pl --list
load_theo_proteotypic_scores.pl --set_tag 'YeastCombNR_20070207_ForwDecoy' --input_file 'proteotypic_peptide.txt'
load_theo_proteotypic_scores.pl --delete_set 'YeastCombNR_20070207_ForwDecoy'
Usage: processFasta.pl -f fasta_file -r regex [ -s (-i) -t 9000 -o output_file ]
Usage: processFasta.pl -f fasta one -f fasta two -f fasta three -m -v -o output_file
-f, --fasta_file Name of fasta input file, required
-o, --output_file Name of output file, defaults to STDOUT
-r, --regex Regular expression applied to defline to define matching
subset
-e, --exclude 'Invert' regex, ie exclude matches instead of non-matches
-m, --merge_files Merges two or more fasta files to a sequence unique
combined file. The first accession encountered for a
given sequence is kept. Does not honor -t, -s, or -r
options.
-v, --verbose Verbose output
-t, --trim_seq Trims the sequence to the specified number of characters.
-h, --help Print this usage and exit
-s, --sprot_extract If set, run sprot extraction to pull accession from pipe
delimited descriptor.
-i, --ipi_extract If set, perform standard 'fix' to descriptor line
Usage: calculateNGenomeMappings.pl -p peptide_file -f protein_fasta_file -o output_file -m map_adaptor
-f, --fasta_file Reference fasta file of proteins for mapping (req)
-p, --peptide_file Merged and sorted file of proteotypic scored
peptides (req)
-o, --output_file Output file name for proteotypic peptides with
mapping (req)
-d, --dbname Ensembl mapping database, see below for version
52 values (req)
-y, --yeast Use Yeast SGD accessions for Ensembl mapping
DB names of 2009-08-15:
bos_taurus_core_52_4b
caenorhabditis_elagans_core_52_190
drosophila_melanogaster_core_52_54a
homo_sapiens_core_52_36n
mus_musculus_core_52_37e
pan_troglodytes_core_52_21j
rattus_norvegicus_core_52_34u
saccharomyces_cerevisiae_core_52_1i
This section outlines the steps taken to create and process the new mouse reference database 2009-08.
Add /regis/sbeams/bin to PATH export PATH=/regis/sbeams/bin/:$PATH
1: Fetch up-to-date data sources, do some light processing.
IPI - version 3.62 (mouse 3.62 56733) wget ftp://ftp.ebi.ac.uk/pub/databases/IPI/current/ipi.MOUSE.fasta.gz wget ftp://ftp.ebi.ac.uk/pub/databases/IPI/current/README -O readme.ipi gunzip ipi.MOUSE.fasta.gz Fix IPI fasta accession line, forces seqs to one line. processFasta.pl -f ipi.MOUSE.fasta -i -v -o mouse_ipi_fixed-acc.fasta
Ensembl ftp://ftp.ensembl.org/pub/current_fasta/mus_musculus/pep/ wget ftp://ftp.ensembl.org/pub/current_fasta/mus_musculus/pep/Mus_musculus.NCBIM37.55.pep.all.fa.gz gunzip Mus_musculus.NCBIM37.55.pep.all.fa.gz wget ftp://ftp.ensembl.org/pub/current_fasta/mus_musculus/pep/README -O readme.ens Streamline the sequence to a single line processFasta.pl -f Mus_musculus.NCBIM37.55.pep.all.fa -v -o mouse_ensembl.fasta
Swiss Prot Fetch and then filter with processFasta, extracting MOUSE entries, fixing accession, main sp wget ftp://ftp.ebi.ac.uk/pub/databases/uniprot/knowledgebase/uniprot_sprot.fasta.gz processFasta.pl -f uniprot_sprot.fasta -v -s -r '_MOUSE' -o mouse_sprot_main.fasta
isoforms file wget ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/complete/uniprot_sprot_varsplic.fasta.gz /regis/sbeams/bin/extractFasta.pl -f uniprot_sprot_varsplic.fasta -r '_MOUSE' -s -o swiss-prot_varsplice_mouse.fasta processFasta.pl -f uniprot_sprot_varsplic.fasta -v -s -r '_MOUSE' -o mouse_sprot_isoforms.fasta
concatenate - use processFasta to eliminate redundancy... processFasta.pl -f mouse_sprot_main.fasta -f mouse_sprot_isoforms.fasta -m -o mouse_sprot_merged.fasta -v
cRAP cp /regis/dbase/users/sbeams/cRAP/crap.fasta . processFasta.pl -f crap.fasta -s -o crap_clean.fasta
Decoys from original search/reference database (not common?) cp /net/db/projects/PeptideAtlas/pipeline/output/Mouse_2008-12_Ens47_P0.9/DATA_FILES/Mus_musculus.fasta ./Original_Mouse_reference_db.fasta processFasta.pl -f Original_Mouse_reference_db.fasta -r 'DECOY_' -v -o mouse_ipi_decoys.fasta
Concatenate all together!
cat mouse_ensembl.fasta mouse_sprot_merged.fasta mouse_ipi_fixed-acc.fasta mouse_ipi_decoys.fasta crap_clean.fasta > mouse_reference_2009-08.fasta
Count unique/redundant seqs by 'merging' file to itself! processFasta.pl -f mouse_reference_2009-08.fasta -m -v -o mouse_reference_non-redundant_2009-08.fasta
total_files => 1 total_seqs => 133420 unique => 76830 redundant => 56590
Finally, for the proteotypic peptide stuff, remove DECOY seqs and trim any long (> 9000) sequences. processFasta.pl -f mouse_reference_2009-08.fasta -t 8999 -r DECOY_ -e -v -o mouse_reference_trimmed_no-decoy.fasta
2: Run the proteotypic scripts. kin => 3238189 no_pdp_prot => 488 no_ps_prot => 819640 orphan => 956 pdp_nan => 3086 pdp_no => 10123 pdp_ok => 3229022 prots => 121195 psieve_cterm_no => 17555 psieve_cterm_ok => 44824 psieve_has_first => 73870 psieve_has_last => 62903 psieve_no => 1091825 psieve_nterm_no => 31861 psieve_nterm_ok => 41755 psieve_ok => 2011325
3239145 merged_proteotypic.tsv
wc: wc: No such file or directory
3239145 total
mergeProteotypicScores.pl -f mouse_reference_nodecoys_2009-08.fasta -p results.PS -d results.PDP -o proteotypic_merged.tsv sortEnsFirst.pl proteotypic_merged.tsv > proteotypic_merged-sorted.tsv
nohup calculateNGenomeMappings.pl -f mouse_reference_nodecoys_2009-08.fasta -p proteotypic_merged-sorted.tsv -o proteotypic_merged-sorted-mapped.tsv -d mus_musculus_core_52_37e &
