MyESL

MyESL implements evolutionary sparse learning using supervised machine learning to build evolutionary models where genomic sites and loci are parameters, connecting phylogenetic hypotheses with sequence variation across genomic loci.

Key Features:

• Bi-level sparsity modeling: Constructs models with logistic regressions using Least Absolute Shrinkage and Selection Operator with bi-level sparsity for phylogenetic trees and alignments.
• Data preprocessing: Processes input phylogenomic alignments for evolutionary sparse learning analyses.
• Result post-processing: Generates molecular evolutionary metrics from output models.
• Computational efficiency: Core algorithms written in C++ with Python-based preprocessing and post-processing tools.
• Cross-platform compatibility: Available as a standalone executable for integration into scripts and third-party software.

Scientific Applications:

• Phylogenetic quality control: Identifies outlier sequences and fragile clades in inferred phylogenies.
• Convergent trait modeling: Builds genetic models of convergent traits from phylogenomic data.

Methodology:

MyESL uses Least Absolute Shrinkage and Selection Operator regression with bi-level sparsity to connect phylogenetic hypotheses with sequence variation across genomic loci. The core implements logistic regressions with bi-level sparsity in C++, with input data preprocessing and result post-processing developed in Python.