Alphabet

Alphabet identifies groups of amino acids that co-occur in columns of aligned protein sequences stored in BLOCKS+ format to define conserved substitution groups, construct eMOTIF sequence motifs, and detect column correlations via MCSeq.

Key Features:

• Column-wise analysis: Examines aligned protein sequences column-by-column to detect groups of amino acids that co-occur more frequently than expected from overall composition.
• Reduced alphabet formation: Creates a reduced alphabet by grouping conserved amino acid substitutions to simplify representation of sequence motifs.
• Empirical conservation identification: Identifies empirically conserved amino acid substitution groups across alignment columns.
• Conditional distribution matrix: Uses a conditional distribution matrix that extends pairwise substitution matrices, with rows representing amino acid distributions conditioned on specific substitution groups.
• Database processing: Converts BLOCKS and HSSP aligned-sequence databases into conditional distribution matrices for systematic examination of substitution groups.
• MCSeq correlation detection: Detects correlations between alignment columns using the MCSeq method.
• eMOTIF construction: Constructs sequence motifs (eMOTIFs) from the reduced alphabet and conserved substitution groups.
• Empirical findings: Has identified twenty conserved amino acid substitution groups in the BLOCKS and HSSP databases.

Scientific Applications:

• Sequence motif construction (eMOTIFs): Supports construction of sequence motifs that aid interpretation of protein function and interaction.
• Column correlation analysis (MCSeq): Enables detection of correlated columns to elucidate structural and functional relationships within proteins.
• Evolutionary and biochemical inference: Provides evidence of conserved biochemical properties and substitution patterns relevant to studies of protein evolution, structure, and function.

Methodology:

Alphabet builds conditional distribution matrices that extend pairwise substitution matrices, represents each conditioning group as a row of amino acid distributions at aligned positions, converts BLOCKS and HSSP data into these matrices, and examines each possible substitution group for empirical conservation, identifying twenty conserved groups.