Using SLiMFinder on Phage Display Data (or other peptides)

Although SLiMFinder is designed with whole protein sequences in mind, it can also be used to identify statistically over-represented motifs in peptide data, including phage display results. Indeed, it is the third example application in the original SLiMFinder paper.

Unfortunately, the SLiMFinder webserver is currently not set up for phage display analysis, so if you are interested in this kind of work then you will need to download SLiMSuite.

Suggested settings for phage display data are below. If anyone has a go and/or wants more advice, please get in touch. (If you try it, I’d be interested to hear how well it works!) Similarly, if you want some advice/ideas on how to combine the peptides with interaction data and full length protein sequences for a more sophisticated analysis, send me a bit more info and I’d be happy to make some suggestions.

Custom settings for phage display data

Here is an overview of the settings that should be tweaked for phage display analysis:

Amino acid frequencies. One thing you will want to try is changing the way that the amino acid frequencies are used. By default, SLiMFinder will use the amino acid frequencies of the input dataset but for phage display peptides this is not really right as the peptides are clearly biased in their composition due to the motifs they contain. Instead, you probably want to set the amino acid frequencies for the background model to those of the human proteome (for human peptides) or even a uniform amino acid distribution. (Select frequencies that model the pre-screening amino acid frequencies.) This is done using the aafreq=FILE option, where FILE can be a fasta file of protein sequences or a delimited file of aa frequencies with the headings “AA” and “FREQ”. (See the manual for details.) If in doubt, try a few runs with different amino acid frequencies.

Evolutionary Filtering. Evolutionary filtering should be switched off (efilter=F) but you will also want to make sure that there is no redundancy in your peptides. (rje_seq.py can be used for this.)

SLiMChance. If you are not so interested in the statistical significance and primarily want to use SLiMFinder to return a ranked list of interesting motifs in the data, set sigcut=1.0 and choose the number of motifs to return with topranks=X.

Ambiguity. Peptide data is usually pretty quick to run, and so it is probably worth exploring the full range of ambiguity with combamb=T (combined amino acid and variable-lengh wildcards). The basic equiv=LIST set for aa degeneracy should be OK for most jobs but you can easily tweak it to add or remove ambiguity combinations as appropriate.

Masking. You will probably want to switch off all masking (masking=F). Low complexity masking might be useful but metmask=F posmask="" should be used as the N-termini are not true protein N-termini.

File management for large SLiMSuite runs

The latest release of SLiMSuite features a slight modification to the way that files are generated and tidied, which can be beneficial for large runs.

Previously, a different results directory (resdir=PATH) was required for each different run to avoid dataset-specific results being over-written. The partial exception was the *.pickle.gz file, which included some SLiMBuild information in its name. (This is predominantly to speed up the ability of (Q)SLiMFinder to recognise when an intermediate pickle file can be used or not.) As of the latest release, the RunID (runid=X) is also now included in dataset-specific output, allowing results from several different runs (with different RunIDs) to go into the same results directory.

The exception is the files that are created as part of the initial setup/SLiMBuild process: *.slimdb, *.dis.tdt and *.upc. From a given Dataset and RunID, the following files will therefore be generated in ResDir/

Dataset.RunID.cloud.txt
Dataset.RunID.mapping.fas
Dataset.RunID.maskaln.fas
Dataset.RunID.masked.fas
Dataset.RunID.motifaln.fas
Dataset.RunID.occ.csv
Dataset.dis.tdt
Dataset.#SLiMBuild-Text#.pickle.gz
Dataset.slimdb
Dataset.upc

Note that the default ResDir is SLiMFinder/, QSLiMFinder/ or SLiMProb and the default RunID is the date and time of the run.

TarGZ and SaveSpace

Obviously, the results directory can quickly fill up with files if there are multiple datasets and/or runs with different RunIDs. The way to get round this is to use the targz=T and savespace=X options.

targz=T will package up all of the files associated with a specific run into a single Dataset.RunID.tgz file. This does not work on Windows. (Note that previous versions generated a Dataset.tar.gz file.) The *.pickle.gz file associated with the run will not be included in the tar file unless savespace=2+ (see below).

Note: the tar file is actually generated from the run directory, not the results directory and will include the relative path to ResDir in the tarred files. This means that if you enter ResDir/ and then tar -xzf Dataset.RunID.tgz, an additional ResDir/ will be created in which the files can be found. This is actually pretty useful as it allows the user to unpack individual runs and then delete the whole directory when finished. To return individual results to their “rightful” place, simply run the tar command from the same directory that the SLiMSuite program was run from (e.g. tar -xzf ResDir/Dataset.RunID.tgz).

The savespace=X option saves space by deleting excess files. It is strongly recommended that this is used in conjunction with the targz=T. There are now four levels of savespace=X:

• 0 = Delete no files
• 1 = Delete all bar *.upc and *.pickle (Pickle excluded from tar.gz with this setting)
• 2 = Delete all bar *.upc files (Pickle included in tar.gz with this setting)
• 3 = Delete all dataset-specific files including *.upc and *.pickle (not *.tar.gz)

Another way to think of this is that 0 will delete nothing, 1 will leave enough files to rerun the same dataset/SLiMBuild combination, 2 will leave enough to run the same dataset with additional SLiMBuild settings, whilst 3 will cleanup absolutely everything.

The recommended setting for running on a cluster or supercomputer is targz=T savespace=1 unless file numbers are an issue, in which case targz=T savespace=2 would be better. targz=T savespace=3 is only really recommended when you are confident that all datasets will run to completion without issues. If there is a chance of nodes going down or walltimes being reached, it is better to keep the pickle files accessible for re-runs.

Using SLiMFinder to discover "local motifs" in protein sequences

The makers of the highly successful MEME Suite have another tool out:

DLocalMotif: A discriminative approach for discovering local motifs in protein sequences

I've not had a chance to go over it in detail but it looks like it could be pretty useful, especially for subcellular targeting motifs. There is one thing that rankles me slightly, though. They define a "local motif" as

"patterns in DNA or protein sequences that occur in a short sequence interval relative to a sequence anchor or landmark."

They then go on to say:

"We believe that DLocalMotif is the only tool for discovering local motifs in protein sequences."

This is just a quick post to point out that SLiMFinder will happily find "local motifs" in protein sequences using the start and end of the sequence as an anchor or landmark. I think it is more limited than DLocalMotif as it is restricted to SLiMs that are very proximal to the sequence termini but it features the usual SLiMChance probability calculations and corrections for evolutionary relationships. (Even without restricting to searches relative to anchor points, SLiMFinder is very successful at finding the KDEL motif and C-terminal PDZ ligand motifs.) The max distance from the termini can be set by maxwild=X up to a limit of 9aa.

If you want to restrict yourself to just N- or C-terminal motifs, use the musthave=LIST option:

musthave="^" for N-terminal motifs.

musthave="$" for C-terminal motifs.

musthave="^,$" for both.

If you want to anchor the motifs internally, this can be done too with a bit of imagination. Just insert an non-standard amino acid character (e.g. Z) at the anchor position, set the expanded alphabet using alphabet=LIST and then force the motif to have the new symbol using musthave=X, e.g.:

alphabet="A,C,D,E,F,G,H,I,K,L,M,N,P,Q,R,S,T,V,W,Y,Z" musthave=Z

I must confess that I have never tried this but it should work and I am happy to help iron out any wrinkles.

(BTW, SLiMFinder also has an experimental feature for using a negative dataset (negatives=FILE if anyone wants to try it out.)

SLiMSuite servers temporarily down

Following some technical issues that have arisen due to some quite major behind-the-scenes re-organisation, the Bioware webserver implementations of SLiMSuite programs have been temporarily taken off-line.

The plan is to give them a bit of attention, iron out any wrinkles that have developed, and then hopefully get them back online over the next week or so. The servers affected are:

• CompariMotif
  
• GOPHER
  
• SLiMDisc (now part of SLiMFinder)
  
• SLiMFinder
  
• SLiMSearch
  
• SLiMSearch 2.0
  

These programs (except SLiMSearch 2.0) are available for download. If you have a particular need for any of these servers, please contact me and I will accelerate its re-appearance.

As each server becomes available, it will be posted here.

SLiMSuite servers and programs

An emerging field of biology is the role of intrinsically disordered regions in protein function and, specifically, protein-protein interactions (PPI) [1-2]. Of particular interest, Short, Linear Motifs (SLiMs) playing a vital role in disorder-mediated PPI, acting as ligands for molecular signalling, post-translational modifications and subcellular targeting [3]. SLiMs have extremely compact protein interaction interfaces, generally encoded by less than 4 major affinity-/specificity-determining residues within a stretch of 2-10 residues [4]. Their small size enables high functional density and evolutionary plasticity, which is frequently exploited by rapidly evolving pathogens that use them to hijack cellular processes [5]. These same features also make experimental discovery a challenge and considerable attention has therefore been given to computational methods for SLiM prediction and analysis [6].

A number of these tools have been developed by the Edwards and Shields labs [7-11] and made available as part of the SLiMSuite package and online as webservers (http://bioware.ucd.ie) [9-10,12-14], with two new tools, SLiMPrints and QSLiMFinder, currently in preparation for submission, and SLiMMaker to be added soon. The main tools that form the SLiMSuite package/servers are as follows:

• SLiMFinder [8,13]: de novo SLiM prediction based on a statistical model of over-represented motifs in unrelated proteins.
  
• SLiMDisc [7,12]: de novo SLiM prediction based on heuristic ranking of over-represented motifs in unrelated proteins.
  
• SLiMPred [11]: de novo SLiM/MoRF prediction in single proteins based machine learning of motif attributes.
  
• SLiMSearch [10]: biological context (disorder & conservation) for searches of pre-defined motifs with under- and over-representation statistics, correcting for evolutionary relationships.
  
• SLiMSearch 2.0 [14]: biological context (disorder & conservation) and ranking for proteome-wide searches of pre-defined motifs.
  
• SLiMPrints (in prep.): de novo SLiM/MoRF prediction in single proteins from statistical clustering of conserved disordered residues.
  
• QSLiMFinder (server coming soon): Query-based variant of SLiMFinder with increased sensitivity and specificity.
  
• CompariMotif [9]: Motif-motif comparison tool.
  
• SLiMMaker (coming soon): Simple tool for converting aligned peptides or SLiM occurrences into a regular expression motif.
  
• GOPHER [12]: Automated orthologue prediction and alignment algorithm. Used for conservation-based masking (SLiMFinder/SLiMSearch) and prediction (SLiMPrints).
  
• GABLAM [7] (server coming soon): BLAST-based protein similarity scoring and clustering. Used for SLiMFinder and SLiMSearch adjustments for evolutionary relationships.
  

Personnel (and funding applications) permitting, a number of improvements for these resources are planned, including updates to the underlying databases for proteome-wide predictions (SLiMSearch 1.0 & 2.0), conservation analyses (SLiMSearch 1.0 & 2.0, SLiMPrints, GOPHER) and SLiM comparisons (CompariMotif). We also intend to improve the integration of different tools, allowing seamless continuation of analyses. Motif predictions ((Q)SLiMFinder/SLiMPrints/SLiMPred) will be able to be searched directly against known motifs (CompariMotif) or proteomes (SLiMSearch); GOPHER alignments will be accessible for SLiMPrints analyses and even SLiMSearch/(Q)SLiMFinder input; outputs of motif occurrences ((Q)SLiMFinder/SLiMSearch) can be used to redefine motifs using SLiMMaker etc. If you have any other suggestions for improvements, please let us know.


References:

[1] Tompa P (2011) Unstructural biology coming of age. Curr Opin Struct Biol 21: 419; [2] Babu MM et al. (2011) Intrinsically disordered proteins: regulation and disease. Curr Opin Struct Biol 21:432; [3] Diella F et al. (2008) Understanding eukaryotic linear motifs and their role in cell signaling and regulation. Front Biosci 13:6580; [4] Davey NE et al. (2012) Attributes of short linear motifs. Mol Biosyst 8:268; [5] Davey NE, Trave G & Gibson TJ (2011) How viruses hijack cell regulation. Trends Biochem Sci 36:159; [6] Davey NE, Edwards RJ & Shields DC (2010) Computational identification and analysis of protein short linear motifs. Front Biosci 15:801; [7] Davey NE, Shields DC & Edwards RJ (2006): SLiMDisc: short, linear motif discovery, correcting for common evolutionary descent. Nucleic Acids Res. 34:3546; [8] Edwards RJ, Davey NE & Shields DC (2007): SLiMFinder: A probabilistic method for identifying over-represented, convergently evolved, short linear motifs in proteins. PLoS ONE 2:e967; [9] Edwards RJ, Davey NE & Shields DC (2008): CompariMotif: Quick and easy comparisons of sequence motifs. Bioinformatics 24:1307; [10] Davey NE et al. (2010): SLiMSearch: a webserver for finding novel occurrences of short linear motifs in proteins, incorporating sequence context. Lecture Notes in Bioinformatics 6282:50; [11] Mooney C et al. (2012): Prediction of short linear protein binding regions. J Mol Biol 415:193; [12] Davey NE, Edwards RJ & Shields DC (2007): The SLiMDisc server: short, linear motif discovery in proteins. Nuc Acids Res 35:W455; [13] Davey NE et al. (2010): SLiMFinder: a web server to find novel, significantly over-represented, short protein motifs. Nuc Acids Res 38:W534; [14] Davey NE et al. (2011): SLiMSearch 2.0: biological context for short linear motifs in proteins. Nuc Acids Res 39:W56.