Weighting aligned protein or nucleic acid sequences to correct for unequal representation

Area of Science:

• Bioinformatics
• Computational Biology
• Molecular Evolution

Background:

• Sequence databases exhibit significant bias, overrepresenting certain organisms and underrepresenting the full diversity of protein families.
• Existing sequence alignments and profiles may not accurately reflect the complete evolutionary landscape of protein families.
• This skewed representation poses challenges for identifying distantly related proteins using computational methods.

Purpose of the Study:

• To address the limitations of biased sequence databases in protein family analysis.
• To develop and present an algorithm for effectively weighting individual sequences within protein families.
• To enhance the accuracy of database searches for distantly related family members.

Main Methods:

• Development of a novel sequence weighting algorithm.
• Application of the algorithm to protein families, using haemoglobins as an example.
• Evaluation of the algorithm's efficacy in correcting for unequal sequence representation.

Main Results:

• The proposed algorithm effectively assigns weights to individual sequences, accounting for database biases.
• Weighted sequence sets provide a more representative view of protein families compared to unweighted sets.
• Demonstrated improvement in the identification of distantly related proteins through database searches.

Conclusions:

• Sequence weighting is crucial for accurate analysis of protein families from skewed databases.
• The developed algorithm offers a robust solution for correcting representation bias in sequence data.
• This approach enhances the utility of sequence alignments and profiles for broader evolutionary and functional studies.