Alignment I/O - Usage Guide

Overview

This skill handles reading, writing, and converting multiple sequence alignment (MSA) files. It uses Biopython's AlignIO module which provides a consistent interface for many alignment formats used in phylogenetics and sequence analysis.

Prerequisites

pip install biopython

# Optional for Pfam-scale streaming and A2M/A3M handling:
pip install pyhmmer

Quick Start

Tell your AI agent what you want to do: - "Read this Clustal alignment file and show me the sequences" - "Convert my PHYLIP alignment to FASTA format" - "Parse the Stockholm file and extract sequence IDs"

Example Prompts

Reading Alignments

"Load the alignment from alignment.aln and tell me how many sequences it has"

"Parse all alignments from this Stockholm file"

"Read the PHYLIP file and show me the first 50 columns"

Writing Alignments

"Save this alignment as a FASTA file"

"Write the alignment to Nexus format for MrBayes"

"Export to PHYLIP-relaxed format to preserve long sequence names"

Format Conversion

"Convert my Clustal alignment to PHYLIP format"

"Convert all .aln files in this directory to FASTA"

"Change the alignment from Stockholm to Nexus"

Working with Alignment Data

"Extract sequences 1-10 from the alignment"

"Trim the alignment to columns 50-150"

"Get all sequence IDs from the alignment"

What the Agent Will Do

1. Load alignment file using appropriate format parser
2. Access sequences, IDs, and alignment properties
3. Perform requested operations (slice, filter, convert)
4. Write output in specified format
5. Report alignment statistics (length, sequence count)

Format Selection for Downstream Tools

Tips

• Always use phylip-relaxed over phylip unless the downstream tool explicitly requires strict format. Strict PHYLIP truncates names to 10 characters, which can silently merge distinct sequences with shared prefixes
• Stockholm format preserves annotations (secondary structure, per-residue quality, metadata) that all other formats lose. Keep a Stockholm master copy if annotations matter
• PHYLIP has two incompatible layout variants (interleaved vs sequential); tools expecting one will fail silently on the other. Use phylip-sequential for PAML
• FASTA is the most portable but carries zero annotation information
• For large alignments, parse() is more memory efficient than read()
• When converting, check what metadata the target format supports. Stockholm-to-FASTA conversion silently discards all annotations