GenDBWiki/CoreDocumentation/FunctionPrediction
Tools for function prediction integrated in GenDB
In this section tools are described that can be used to predict the function of functional DNA regions like protein coding genes (CDS) within the GenDB annotation system. These tools are autmatically configured by the installation system.
Currently integrated programs
| Program | Task | Reference |
| BLAST | Search for sequence similarities | (http://nar.oupjournals.org/cgi/content/full/25/17/3389) |
| HMMER | Prediction of protein families and domains | (http://hmmer.wustl.edu/) |
| InterPro | Prediction of protein function and structure | (http://nar.oupjournals.org/cgi/content/full/31/1/315) |
| SignalP | Detection of signal peptide cleavage sites in proteins | SignalP |
| TMHMM | Search for transmembrane regions of proteins | TMHMM |
| Helixturnhelix | Detection of helix-turn-helix nucleic acid binding motifs in proteins | EMBOSS |
Documentation for individual programs
In the following sections you can find detailed descriptions for all programs that can be used to predict region functions within GenDB.
BLAST
The BLAST programs are widely used tools for searching protein and DNA databases for sequence similarities. In GenDB they are the major tools used to generate observations.
Using genome sequences as Blast target
One of the main features of GenDB is the ability to store only the abstracts of tool results as observations instead of putting the whole result (e.g. a full Blast report with alignments) into the database. To recreate the alignment, a method to get a single entry from a Blast-able database has to exist. GenDB currently uses a BioPerl based index or an SRS query for the main genomic databases like EMBL or SwissProt. If you want to Blast some regions versus a genome (probably the genome of a closely related organism), the default methods will fail. Since the genome database it considers is small (approx. 3-8 mb instead of 1-2 gb), GenDB recreates the alignment by rerunning Blast. There's a small performance penality, but our tests revealed 2-3 seconds for the Blast run on several systems, including highend servers, desktop PCs and laptops.
HMMER
Next to Blast and its databases the HMMER tools are the most important analysis tools integrated into GenDB. Profile hidden Markov models (profile HMMs) can be used to do sensitive database searching using statistical descriptions of a sequence family's consensus sequence. HMMER is a implementation of a profile HMM software for protein sequence analysis. hmmpfam is part of the HMMER package. This tool can be used for scanning e.g. the Pfam or TIGRFAM database for homologous protein domains. You need this tool (and most of the HMMER package) to use Pfam in GenDB!
Pfam database
The Pfam protein families database is a large collection of protein families and accurate protein domain definitions. These models are represented as profile hidden Markov that are constructed and searched using HMMER. The parameters of the profile HMMs are derived from multiple alignments of family member sequences. The main application of Pfam in GenDB is to predict the domain organization of proteins.
To use Pfam within GenDB you have to install the HMMER package and to download the Pfam database (Pfam_ls.gz) from the sanger site.
TIGRFAM database
TIGRFAMs is a collection of manually curated protein families consisting of hidden Markov models (HMMs), multiple sequence alignments, commentary, Gene Ontology (GO) assignments, etc. TIGRFAMs contains models of full-length proteins and shorter regions at the levels of superfamilies, subfamilies and equivalogs, where equivalogs are sets of homologous proteins conserved with respect to function since their last common ancestor. TIGRFAMs is thus complementary to Pfam, whose models typically achieve broad coverage across distant homologs but end at the boundaries of conserved structural domains. The TIGRFAMs models are designed to support function annotation of genomes. The database can be searched using hmmpfam from the HMMER package.
To use TIGRFAM within GenDB you have to install the HMMER package and to download the TIGRFAM database from the download site.
HMMFETCH
Instead of fetching single entries from a nSRS server, the HMMER package provides a tool called hmmfetch, which extracts single HMMs from the HMM file. This tool is used similar to the BioPerl indices described above for recomputing the full Pfam result.
InterPro
InterPRO-Scan is a meta tool that combines several well known analysis tools like Blast and hmmpfam and several databases like PROSITE, Pfam, PRINTS, ProDom, SMART and TIGRFAMs to predict protein families, domains and functional sites of proteins. Hits found by InterPRO-Scan are associated to the InterPRO database, which contains descriptions for functional categories. There's also a mapping from InterPRO to GO categories which can be used for assigning functional categories to genes and other regions. More information about Interpro itself can be found at the InterPRO site on the EBI web server.
InterPRO-Scan support has been integrated into GenDB. To use it, you have to download and install InterPRO-Scan, the backend binaries and the data tarball from the download site (approx. 90 mb).
Setting up InterPRO-Scan is beyond the scope of this documentation. Refer to the included README and FAQ.
SignalP
SignalP predicts the presence and location of signal peptide cleavage sites in amino acid sequences. Information about it can be found here. The tool is not freely available therefore you have to send a request to the author in order to receive it.
Helixturnhelix
Helixturnhelix is part of the EMBOSS package and finds helix-turn-helix nucleic acid binding motifs in proteins. To use the program in GenDB you need to install the EMBOSS package from the MRC download site.