What question did this study set out to answer?

This research aims to develop profile-based string kernels for improving protein classification and remote homology detection.

November 8, 2004

Profile-based string kernels for remote homology detection and motif extraction

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial demonstrates improved remote homology detection in protein sequences, suggesting enhanced classification efficiency.

This research aims to develop profile-based string kernels for improving protein classification and remote homology detection.
Utilized probabilistic profiles from the PSI-BLAST algorithm to create position-dependent mutation neighborhoods.
Implemented support vector machines for classification using these string kernels.
Conducted experiments using the SCOP database to evaluate the effectiveness of the approach.
Profile-based string kernels outperformed existing SVM methods in remote homology detection with significant accuracy improvements.
Discriminative sequence motifs extracted corresponded to important structural features in proteins.
Profile kernel results showed comparable performance to cluster kernels, with better scalability for large datasets.

Kuang et al. (2004) studied this question.

View Full Paper

Ask AI

Bookmark