FermatS

FermatS encodes protein sequences of the 20 amino acids using the Fermat spiral and normalized moments of inertia to compare sequences and predict DNA-binding proteins.

Key Features:

• Numerical Representation: Represents protein sequences via global and local positional information derived from Fermat spiral curves and normalized moments of inertia to produce numerical characteristics.
• Predictive Modeling: Uses logistic regression with 5-fold cross-validation to predict DNA-binding proteins and reported improved performance versus DNAbinder, iDNA-prot, DNA-prot and gDNA-prot with F-measure improvements of 0.0069–0.609 and MCC improvements of 0.293–0.898, particularly on unbalanced datasets.
• Application to Biological Data: Applied to similarity analysis of nine ND5 proteins with results consistent with biological evolution theories.

Scientific Applications:

• DNA-binding protein recognition: Improves identification of proteins involved in gene transcription, regulation, replication, repair, recombination, chromatin formation and ribosome assembly.
• Protein sequence comparison and evolutionary analysis: Supports comparison of protein sequences and analysis of evolutionary relationships, as exemplified by ND5 protein similarity analysis.

Methodology:

Constructs numerical characteristics from protein sequences using the Fermat spiral and normalized moments of inertia to encode global and local positional information, and applies logistic regression with 5-fold cross-validation for prediction.