Improving Profile HMMs for Remote Homology Detection
Author Information
Author(s): Bernardes Juliana S, Dávila Alberto MR, Costa Vítor S, Zaverucha Gerson
Primary Institution: Universidade Federal do Rio de Janeiro
Hypothesis
Can structural alignments improve the performance of profile Hidden Markov Models (pHMMs) in detecting remote homologies?
Conclusion
pHMMs derived from structural alignments perform significantly better than those derived from sequence alignments in low-identity regions.
Supporting Evidence
- Structural alignments were shown to improve pHMM performance in low-identity regions.
- The study used multiple alignment tools including CLUSTALW and TCOFFEE for sequence alignments.
- HMMER and SAM were the two pHMM packages compared in the study.
Takeaway
This study shows that using structural information can help computers find similarities between proteins that look very different.
Methodology
The study used leave-one-family-out cross-validation on the SCOP database, comparing pHMMs built from structural and sequence alignments.
Limitations
Sensitivity of structural alignment tools is still low for low-identity regions.
Statistical Information
P-Value
p ≤ 0.02
Statistical Significance
p ≤ 0.02
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website