Welcome to the Eukaryotic Linear Motif (ELM) resource

This computational biology resource mainly focuses on annotation and detection of eukaryotic linear motifs (ELMs) by providing both a repository of annotated motif data and an exploratory tool for motif prediction. ELMs, or short linear motifs (SLiMs), are compact protein interaction sites composed of short stretches of adjacent amino acids. They are enriched in intrinsically disordered regions of the proteome and provide a wide range of functionality to proteins (Davey,2011,Van Roey,2014) They play crucial roles in cell regulation and are also of clinical importance, as aberrant SLiM function has been associated with several diseases and SLiM mimics are often used by pathogens to manipulate their hosts' cellular machinery (Davey,2011, Uyar,2014)

ELM Prediction

The ELM prediction tool scans user-submitted protein sequences for matches to the regular expressions defined in ELM. Distinction is made between matches that correspond to experimentally validated motif instances already curated in the ELM database and matches that correspond to putative motifs based on the sequence. Since SLiMs are short and degenerate, overprediction is likely and many putative SLiMs will be false positives. However, predictive power is improved by using additional filters based on contextual information, including taxonomy, cellular compartment, evolutionary conservation and structural features.

ELM DB

The ELM relational database stores different types of data about experimentally validated SLiMs that are manually curated from the literature. ELM instances are classified by motif type, functional site and ELM class. A functional site contains one to many ELM classes, which are described by a regular expression and list experimentally validated motif instances matching this sequence pattern. All data curated in ELM DB can be searched on the ELM website according to the following categories:

• 317 annotated ELM classes
• 3,934 experimentally validated ELM instances in 260 taxons
• 142 ELM methods described in 3,867 articles to experimentally validate ELM instances
• 616 solved PDB structures for curated ELM instances (from PDB)
• 156 globular ELM binding domains (from Pfam, SMART, and InterPro)
• 2,394 interactions mediated by curated ELM instances
• 887 regulatory switches mediated by curated ELM instances (from Switches.ELM DB)
• 1,063 pathways from KEGG involving linear motifs annotated in 1297 Sequences
• 277 viral instances interfering with host cellular processes
• 11 ELM related diseases annotated as being caused by aberrant motif function
• 2 examples where pathogens abuse motifs to deregulate host cells

ELM Candidates

The ELM candidates pages contain lists of candidate classes and instances awaiting curation, and can be extended by users: Anybody can submit candidates, please provide as much information as possible. If you're interested in annotating a full ELM entry or individual instances, get in contact with the ELM team for procedures and awards!

ELM Information

The ELM information pages provide additional details to assist users in searching data in the ELM DB and in searching for putative motifs in query sequences. The Help page explains the use of regular expressions to define sequence patterns of ELM classes and to detect putative motifs in user-submitted query sequences, describes the filters that are applied to increase the reliability of the prediction tool, and defines terms frequently used in the ELM resource. The News lists changes and updates made to ELM, while the Links page lists links to other interesting bioinformatics resources. Funding information, participating groups and a list of ELM-related publications can be found on the About page.

ELM Downloads

Data curated in ELM DB can be downloaded and distributed for non-commercial use according to the ELM Software License Agreement.

ELM API

The ELM prediction tool can also be accessed programmatically via the ELM API. Read the API manual for more information.

Disclaimer

Short patterns applied to proteins are usually not statistically significant: Therefore we can't provide E-values as with BLAST searches. This means that most matches shown are more likely to be false positives than true matches. We hope that ELM server results will prove useful as guides to experimentation but they should not be treated as factual findings.