HMM-ModE – Improved classification using profile hidden Markov models by optimising the discrimination threshold and modifying emission probabilities with negative training sequences
2007

HMM-ModE: Improved Protein Classification Method

Sample size: 66 publication 10 minutes Evidence: high

Author Information

Author(s): Prashant K. Srivastava, Dhwani K. Desai, Soumyadeep Nandi, Andrew M. Lynn

Primary Institution: Jawaharlal Nehru University, New Delhi, India

Hypothesis

Can optimizing discrimination thresholds and modifying emission probabilities with negative training sequences improve the classification of protein families using profile hidden Markov models?

Conclusion

The HMM-ModE protocol significantly improves the specificity of protein classification based on molecular function using pre-classified training data.

Supporting Evidence

  • The specificity improved from an average of 21% to 98% after optimization.
  • The method was validated on sequences from six sub-families of the AGC family of kinases.
  • HMM-ModE showed better performance compared to traditional methods in classifying protein kinases.

Takeaway

This study created a new way to classify proteins by using special training data to make better guesses about what proteins do, helping scientists understand them better.

Methodology

The study used profile hidden Markov models (HMMs) with optimized thresholds and modified emission probabilities based on negative training sequences to classify protein families.

Potential Biases

Potential for misclassification if the training data does not adequately represent the diversity of sequences.

Limitations

The method may reduce sensitivity in some cases, particularly when sequences are shorter than the average.

Participant Demographics

The study focused on protein sequences from various sub-families of the AGC family of kinases and G-protein coupled receptors.

Statistical Information

P-Value

p<0.05

Statistical Significance

p<0.05

Digital Object Identifier (DOI)

10.1186/1471-2105-8-104

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication