Sequence Statistics - Usage Guide

Overview

This skill enables AI agents to help you calculate decision-grade statistics for sequence datasets and genome assemblies using Biopython: N50/L50, auN, NG50, length distributions, and GC content with explicit ambiguity handling.

Prerequisites

pip install biopython

Quick Start

Tell your AI agent what you want to do:

"Calculate N50, L50, and auN for my assembly"
"Compute NG50 against a genome size of 3.1 Gb"
"Generate a summary report for these sequences"
"Compare contiguity across multiple FASTA files"
"Report GC content but exclude ambiguous bases"

Example Prompts

Assembly QC

"Calculate N50, L50, auN, and total bases for my genome assembly"

Cross-Assembly Comparison

"Compare NG50 and auN for all FASTA files in the assemblies folder using a 3.1 Gb genome size"

GC Content

"Report mean GC content, excluding N bases, for my contigs"

Distribution

"Show me the contig length distribution histogram for my sequences"

Key Metrics

Metric	What It Tells You
N50	Contiguity using assembly size (not comparable across assemblies)
L50	How many contigs contain half the assembled bases
auN	Smooth, threshold-free contiguity; preferred for ranking
NG50	Contiguity using genome size; comparable across assemblies
NGA50	Like NG50 but broken at misassembly points (NGA50 << NG50 = misassemblies)
GC%	G+C proportion; depends on the chosen ambiguity mode

What the Agent Will Do

Load sequences from the input file (gzip-aware)
Compute length, GC, and N-content arrays in one pass
Derive N50/L50, auN, and (with a genome size) NG50
Report a summary or a side-by-side comparison table

Tips

N50 measures contiguity, not correctness: a misassembly can post a large N50, so pair it with BUSCO completeness and read-backed validation.
Prefer auN for comparing assemblies: it is smooth and every join raises it, whereas N50 can ignore real joins.
Use NG50 (genome-size denominator), not N50, to compare assemblies of the same genome on a common baseline.
Set gc_fraction(seq, ambiguous=...) explicitly: remove excludes Ns, ignore counts them in the denominator, weighted apportions expected GC; the modes give different answers on the same sequence.
gc_fraction returns a fraction 0-1; multiply by 100 only for display (the removed GC() returned a percent).
Median contig length is near-useless for assemblies (dominated by tiny contigs); rely on N50/auN instead.
GC content varies by organism (humans ~41%, E. coli ~51%).

read-sequences - Parse sequences for statistics calculation
batch-processing - Calculate stats across multiple files
fastq-quality - Quality score statistics for FASTQ files
sequence-manipulation/sequence-properties - Per-sequence GC content and properties
alignment-files/bam-statistics - samtools stats/flagstat for alignment statistics