SH3PepInt is an alignment-free, fast and sophisticated graph kernel based tool to predict SH3-peptide interactions. Total 69 built-in models are available for SH3-peptide predictions. It does not need any pre-alignment of the peptides. It computes a window analysis on the query proteins and considers 15 amino acids length peptides. User can change the step size for the window analysis. Depending on the user requirement it uses Gene Ontology database for getting more reliable interactions.

• Kousik Kundu, Martin Mann, Fabrizio Costa, and Rolf Backofen
  MoDPepInt: An interactive webserver for prediction of modular domain-peptide interactions
  Bioinformatics, 2014, in press.
  
• Kousik Kundu, Fabrizio Costa, and Rolf Backofen
  A graph kernel approach for alignment-free domain-peptide interaction prediction with an application to human SH3 domains
  Bioinformatics, 29(13), pp. i335-i343, 2013.
  

Results are computed with SH3PepInt version 1.0.0

PDZPepInt accepts input in form of a multiple FASTA file. An example looks like this:

	
>PIK3C2B
MSSTQGNGEHWKSLESVGISRKELAMAEALQMEYDALSRLRHDKEENRAKQNADPSLISW
DEPGVDFYSKPAGRRTDLKLLRGLSGSDPTLNYNSLSPQEGPPNHSTSQGPQPGSDPWPK

>PPP3CA
MSEPKAIDPKLSTTDRVVKAVPFPPSHRLTAKEVFDNDGKPRVDILKAHLMKEGRLEESV
ALRIITEGASILRQEKNLLDIDAPVTVCGDIHGQFFDLMKLFEVGGSPANTRYLFLGDYV

>SYNJ1
MAFSKGFRIYHKLDPPPFSLIVETRHKEECLMFESGAVAVLSSAEKEAIKGTYSKVLDAY
GLLGVLRLNLGDTMLHYLVLVTGCMSVGKIQESEVFRVTSTEFISLRIDSSDEDRISEVR

Input can be given either as direct text input or uploading a file.
Note: Input size is limited to restrict computation time and memory requirements.

The parameter constraints are: The input has to be in valid FASTA format. The number of sequences has to be at least 0 and at most 500. Sequence lengths have to be in the range 1-5000. The allowed sequence alphabet is 'GPAVLIMCFYWHKRQNEDST'. Either FASTA input or a UniProt ID list have to be provided. In case an enabled filter requires UniProt IDs: Each FASTA sequence header/name has to be just a valid UniProt ID or a valid UniProt FASTA header.
Defaults to ()

Instead of feeding directly protein sequences, you can provide UniProt IDs of the targeted proteins as an input. In this case, the protein sequence will be automatically downloaded from the UniProt database. Multi UniProt IDs separated by below mentioned separators are also accepted. An example looks like this:

 O00750, Q08209, O43426

The parameter constraints are: Has to be a list of 0-500 UniProt IDs that are separated by ([,\.;: \t\n]|\r\n). Access to the UniProt database is needed. The value has to match against the regular expression '^[^;'"]*$'.
Defaults to ()

List of all SH3 protein domains available for an interaction screening.

The parameter constraints are: The value has to match against the regular expression '^[^;'"]*$'. Only protein domains from the following list are allowed : ABL, AMPHIPHYSIN_I, AMPHIPHYSIN_II, ARGBP1, ARGBP2a-1, ARGBP2a-2, ARGBP2a-3, ARHGAP12, ARHGEF16, BOG25, BTK, CAP-1, CAP-2, CIN85-1, CIN85-3, COOL1, CSK, DDEF2, DLG1, DOCK1, ENDOPHB1, ENDOPHILIN1, ENDOPHILIN2, ENDOPHILIN3, EPS8, EPS8L2, FGR, FISH, FNBP1L, FRK, FYN, GRAP2-1, GRAP2-2, GRB2-1, GRB2-2, HCK, INTERSECTIN1-1, INTERSECTIN1-2, INTERSECTIN1-3, INTERSECTIN1-4, INTERSECTIN1-5, IRSP53, LCK, LYN, MLK3, MPP1, MYO7A, NCF1, NCK2, NPHP1, OSTF1, P51NOX, PAC2, PAC3, PIK3R1, PLCG1, RASGAP, RIMB1-2, RIMB1-3, RUSC1, SH3PX3, SNX18, SNX9, SRC, STAM1, STAM2, TUBA-1, TUBA-3, TUBA-6.
Defaults to ()

For retrieving the peptide sequences, the query proteins will be scanned automatically with a window size of 15 and the user defined step size.

The parameter constraints are: Input value has to be parsable as Integer. The value must be greater than or equal to 1.
Defaults to (5)

This filter has been introduced to find the potential proline-rich binding motifs using 31 regular expressions that are most common to describe most of the SH3 domain binding specificity (see also this article).

The parameter constraints are: Input value has to be parsable as Boolean.
Defaults to (true)

This filter was implemented considering the terms relative to the sub-cellular localization hierarchy in the controlled vocabulary of the Gene Ontology database (January, 2013). In case of multiple cellular locations, we consider a peptide viable for interaction if it shares at least one of the terms with the domain. This filter is not applicable for the domains without annotated sub-cellular localizations. Uniprot Ids of the query proteins are mandatory for using this filter: within FASTA provide either ONLY the UniProt ID within the FASTA header OR use the header encoding from the UniProt database; alternatively just provide the UniProt IDs for an automated download of the query sequences.

The parameter constraints are: Input value has to be parsable as Boolean.
Defaults to (false)

This is an example for the interaction prediction of human SH3 domains from ABL, LCK and PIK3R1 proteins with the target proteins PIK3C2B, PPP3CA and SYNJ1.

The example's result can be directly accessed here