scikit-bio

To compute alpha and beta diversity, you process a BIOM feature table with a phylogenetic tree to calculate ecological diversity statistics, including phylogenetic measures like UniFrac, yielding distance matrices for downstream ordination.

Running PERMANOVA distance tests on microbiome datasets requires a symmetric distance matrix and applies permutation-based testing to evaluate statistical significance, utilizing efficient reading and generators to handle large biological datasets.

Yes, you can perform multiple sequence alignment and phylogenetic tree construction directly from FASTA files by computing local and global alignments, then building trees from distance matrices using Newick I/O support.

Yes, this approach supports ordination methods like PCoA and CCA for microbial ecology by computing principal coordinates analysis, canonical correspondence analysis, and redundancy analysis directly on generated distance matrices.

Reading and writing biological sequence formats like FASTQ and GenBank is handled through standard file I/O, utilizing efficient streaming and generators to process large datasets without loading entire files into memory.

Limitations when applying protein embeddings to downstream machine learning workflows include ensuring ID-consistent inputs and proper alignment with feature tables, as this processing focuses on biological data workflows rather than generating the embeddings themselves.