SLiMMine is an online resource for discovering and analyzing short linear motifs (SLiMs) in human proteins. It is largely based on the regular expressions defined in the Eukaryotic Linear Motif (ELM) resource.

Because SLiMs are degenerate, scanning protein sequences with regular expressions alone produces a high number of false positives. To address this, SLiMMine uses a neural network to predict motif probabilities from protein embeddings, filtering out approximately 80% of likely false-positive hits. In addition, the annotation of each ELM class has been refined with stringent Gene Ontology terms and binding partner information to further exclude motifs occurring in unfavorable biological contexts.

The SLiMMine resource contains 696,329 SLiMs derived from ELM classes, as well as 32,501 newly discovered motifs.

The platform can be used in three ways:

1. Search for any human protein to analyze predicted SLiMs
2. Search for any ELM class to identify human proteins that contain the given motif
3. Search for any regular expression

For more information see the manual.

If you used this resource for your work, please cite:
Reliable prediction of short linear motifs in the human proteome
Rita Pancsa, Erzsébet Fichó, Zsófia E. Kálmán, Csongor Gerdán, István Reményi, András Zeke, Gábor E. Tusnády, Laszlo Dobson
in preparation