QuasiMotiFinder: protein annotation by searching for evolutionarily conserved motif-like patterns

Area of Science:

• Bioinformatics
• Computational Biology
• Protein Science

Background:

• Protein annotation relies on sequence signature databases like PROSITE, but accuracy issues like false positives and negatives persist.
• Traditional methods use simple pattern searches, which can be unreliable for predicting protein function.
• Emerging approaches utilize sequence profiles derived from multiple sequence alignments (MSA) for more accurate annotation.

Purpose of the Study:

• To introduce QuasiMotiFinder, a novel web server for protein annotation.
• To evaluate QuasiMotiFinder's performance in reducing prediction errors compared to existing methods.
• To offer a complementary approach to profile-based searches for enhanced protein function identification.

Main Methods:

• Developed QuasiMotiFinder, a web server implementing a novel search strategy.
• Employed MSA of homologous query proteins searched against original PROSITE signatures.
• Incorporated average evolutionary conservation of signatures into the search algorithm.

Main Results:

• QuasiMotiFinder significantly reduces false positive predictions compared to simple pattern searches.
• The server demonstrates a reduced rate of false negative predictions by allowing for physicochemically similar patterns.
• QuasiMotiFinder's performance is comparable to profile-based search methods.

Conclusions:

• QuasiMotiFinder offers an effective alternative for protein annotation, improving accuracy.
• The method complements existing profile search approaches, providing a more comprehensive annotation strategy.
• Accurate protein function prediction is crucial for various biological and biomedical applications.