Afann

Afann adjusts alignment-free sequence dissimilarity measures using neural network regression to correct bias and enable accurate comparison of genome sequences or raw sequencing samples without assembly.

Key Features:

• Alignment-Free Comparison: Rapid calculation of dissimilarity measures including d2star, d2shepp, CVtree, Manhattan, Euclidean, and d2 for genome sequences or raw sequencing samples without assembly.
• Bias Adjustment: Correction of biases inherent in alignment-free dissimilarity calculations derived from sequencing data to mitigate overestimation when comparing sequencing samples directly rather than assembled genomes.
• Neural Network Regression: Use of neural network regression to refine and correct calculated dissimilarities for improved accuracy.
• Computational Efficiency: Time-efficient and memory-conservative operation suitable for large-scale sequence comparison.

Scientific Applications:

• Comparative Genomics: Provides bias-adjusted dissimilarity measures to support accurate comparisons between genomes and within populations.
• Microbiome Analysis: Enables analysis of complex microbiome sequencing samples without assembly, reducing computational burden from alignment-based methods.
• Evolutionary Studies: Facilitates exploration of evolutionary relationships and genetic diversity using alignment-free metrics.

Methodology:

Calculate dissimilarity measures from sequencing samples using alignment-free methods, then apply a neural network regression model to adjust those measures and correct biases that cause overestimation relative to genome-based calculations.