quantiprot

quantiprot computes quantitative characterizations of protein sequences to define multidimensional feature spaces for alignment-free similarity analysis, comparative studies across protein families and organisms, and evaluation of generative sequence models.

Key Features:

• Implementation: Implemented in Python for computational analysis of protein sequences.
• Multidimensional solution space: Defines a multidimensional feature space based on quantitative properties to represent sequence relationships independently of alignments.
• Characteristic calculations: Calculates characteristics directly from amino acid sequences or using physico-chemical properties of amino acids.
• Recurrence and determinism analysis: Performs quantitative analysis of recurrence and determinism within sequences.
• n-gram distributions: Computes n-gram distributions to characterize sequence patterns.
• Zipf's law coefficient: Computes Zipf's law coefficient for analysis of statistical distributions of sequence elements.
• Alignment-free similarity searches and clustering: Enables alignment-free similarity searches and clustering of large or divergent protein sequence sets.
• Comparative feature-space analyses: Facilitates comparative studies across protein families and organisms by mapping sequences into the defined feature space.
• Generative model evaluation: Supports evaluation of generative models by comparing generated sequences with observed sequences in the feature space.

Scientific Applications:

• Alignment-free similarity search and clustering: Identification and clustering of similar protein sequences without relying on sequence alignments.
• Protein family analysis: Comparative analysis of evolutionary dynamics and functional conservation within protein families using quantitative features.
• Organismal comparative genomics: Comparison of proteomes across organisms to investigate biological diversity and evolutionary relationships.
• Model evaluation and benchmarking: Assessment of generative sequence models by comparing generated sequences to observed sequences in quantitative feature space.

Methodology:

Extracts sequence-derived and amino-acid physico-chemical properties, calculates recurrence and determinism metrics, computes n-gram distributions and Zipf's law coefficients, and maps sequences into a multidimensional quantitative feature space for alignment-free similarity searches, clustering, comparative analyses, and comparison of generated versus observed sequences.