Cluster DNA sequences using Fourier power spectrum

Cluster DNA sequences using Fourier power spectrum clusters DNA sequences by converting nucleotide sequences into binary indicator sequences, computing Discrete Fourier Transform (DFT) power spectra, extracting mathematical moments to form multidimensional feature vectors, and applying cluster analysis to classify genes and genomes and infer evolutionary relationships.

Key Features:

• Binary indicator sequences: Constructs a binary indicator sequence for each nucleotide type (A, T, C, G) from each input DNA sequence.
• Discrete Fourier Transform (DFT): Applies DFT to the binary indicator sequences to compute power spectra as frequency-domain representations.
• Mathematical moments: Calculates mathematical moments from the power spectra to capture statistical properties and produce multidimensional real-valued vectors.
• Cluster analysis: Performs cluster analysis on the resulting vectors to determine evolutionary relationships and classify sequences without alignment.
• Alignment-free comparison: Enables comparison of sequences without sequence alignment, suitable for sequences of different lengths.
• Scalability and speed: Provides faster processing than conventional multiple sequence alignment methods, facilitating large-scale genomic analyses.

Scientific Applications:

• Gene and genome classification: Classifies genes and genomes using spectral-feature-based clustering to provide taxonomic and functional grouping.
• Evolutionary relationship inference: Infers evolutionary relationships among DNA sequences using alignment-free cluster analysis of spectral features.
• Large-scale comparative genomics: Applies to large datasets where faster, alignment-free comparisons are required for genomic studies.
• Analysis of variable-length sequences: Suited for comparative analyses when input sequences have differing lengths or when alignment is problematic.

Methodology:

Transform each DNA sequence into binary indicator sequences for A, T, C, G; apply Discrete Fourier Transform to obtain power spectra; compute mathematical moments from the spectra to form multidimensional real-valued vectors; perform cluster analysis on those vectors.