Partitioning and correlating subgroup characteristics from Aligned Pattern Clusters.

Area of Science:

• Bioinformatics
• Computational Biology
• Protein Sequence Analysis

Background:

• Evolutionarily conserved amino acids highlight functional protein regions, while variations indicate divergence.
• Existing methods struggle with erroneous prior knowledge or sequence dissimilarities like insertions/deletions.
• Supervised algorithms are hampered by unavailable or incorrect biological function and species data.

Purpose of the Study:

• Develop a novel unsupervised algorithm for discovering highly conserved protein regions.
• Overcome limitations of current methods in handling sequence divergence and lack of prior biological information.
• Identify evolutionary patterns and functional characteristics within protein families.

Main Methods:

• Developed a novel unsupervised algorithm utilizing data measures from input sequences.
• Incorporated class measures based on a priori groupings to reveal subgroups and functional characteristics.
• Applied the algorithm to uncharacterized protein families to group related sequences and identify conserved regions (Aligned Pattern Clusters).

Main Results:

• The algorithm successfully grouped evolutionarily related sequences and identified conserved regions within and across proteins.
• Demonstrated that data measures are unbiased and class measures accurately rank evolutionary groupings.
• Combined measures allowed inference of biologically important regions within protein binding domains.
• Achieved superior runtime and comparable accuracy to popular supervised methods.

Conclusions:

• The novel unsupervised algorithm effectively identifies conserved protein regions and evolutionary patterns.
• The method provides a robust approach for analyzing divergent protein sequences without relying on potentially erroneous prior knowledge.
• This tool aids in inferring functional characteristics and biological importance of protein regions, particularly in uncharacterized families.