G4Catchall: A G-quadruplex prediction approach considering atypical features.

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• G-quadruplexes are crucial DNA and RNA structures, but their in vivo discovery is challenging due to size, topology, and environmental factors.
• In vitro methods cannot fully replicate in vivo conditions and are impractical for large sequences.
• Current computational prediction tools often miss G-quadruplexes due to unconventional features like disrupted G-tracts or long loops.

Purpose of the Study:

• To develop a novel computational tool for more accurate G-quadruplex discovery.
• To improve G-quadruplex prediction by incorporating features missed by existing methods.
• To enhance the understanding of G-quadruplex formation in genomic DNA and RNA.

Main Methods:

• Developed a novel computational approach for G-quadruplex discovery.
• Incorporated features of previously overlooked G-quadruplex-forming sequences.
• Validated the tool against experimentally confirmed G-quadruplex sequences.

Main Results:

• Achieved a sensitivity of 99% and Youden's J-statistics of 0.91 in predicting G-quadruplexes.
• Demonstrated improved accuracy compared to other computational approaches.
• Showed that allowing atypical G-tracts and extreme loop sizes benefits prediction accuracy.

Conclusions:

• The novel tool offers a significant improvement in G-quadruplex discovery accuracy.
• The tool's ability to account for unconventional sequence features enhances its utility.
• This advancement provides a more reliable method for identifying G-quadruplexes in genomic contexts.