菜单
Track bacterial strains using MASH, sourmash, fastANI, and inStrain. Compare genomes, detect contamination, and monitor strain-level variation. Use when needing sub-species resolution for outbreak tracking, transmission analysis, or within-host strain dynamics.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/zja2004/BGI-CLI --skill bio-metagenomics-strain-tracking命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Based on the Recursive Language Models (RLM) research by Zhang, Kraska, and Khattab (2025), this skill provides strategies for handling tasks that exceed comfortable context limits through programmatic decomposition and recursive self-invocation. Triggers on phrases like "analyze all files", "process this large document", "aggregate information from", "search across the codebase", or tasks involving 10+ files or 50k+ tokens.
This skill should be used when the user asks to "design agent tools", "create tool descriptions", "reduce tool complexity", "implement MCP tools", or mentions tool consolidation, architectural reduction, tool naming conventions, or agent-tool interfaces.
This skill should be used when the user asks to "start an LLM project", "design batch pipeline", "evaluate task-model fit", "structure agent project", or mentions pipeline architecture, agent-assisted development, cost estimation, or choosing between LLM and traditional approaches.
正在显示 SKILL.md
基于 SOC 职业分类
| name | bio-metagenomics-strain-tracking |
| description | Track bacterial strains using MASH, sourmash, fastANI, and inStrain. Compare genomes, detect contamination, and monitor strain-level variation. Use when needing sub-species resolution for outbreak tracking, transmission analysis, or within-host strain dynamics. |
| tool_type | cli |
| primary_tool | MASH |
Reference examples tested with: Bowtie2 2.5.3+, MetaPhlAn 4.1+, numpy 1.26+, pandas 2.2+, samtools 1.19+, scipy 1.12+
Before using code patterns, verify installed versions match. If versions differ:
pip show <package> then help(module.function) to check signatures<tool> --version then <tool> --help to confirm flagsIf code throws ImportError, AttributeError, or TypeError, introspect the installed package and adapt the example to match the actual API rather than retrying.
"Track bacterial strains across my samples" → Resolve sub-species variation using genome sketching (Mash/sourmash), average nucleotide identity (fastANI), or within-sample strain profiling (inStrain) for outbreak tracking and transmission analysis.
mash dist, sourmash compare, fastANI, inStrain profileIdentify and track bacterial strains at sub-species resolution.
| Tool | Method | Best For |
|---|---|---|
| MASH | MinHash sketches | Fast distance estimation |
| sourmash | MinHash + containment | Metagenome comparisons |
| fastANI | ANI calculation | Accurate species/strain ID |
| inStrain | SNV profiling | Strain dynamics in metagenomes |
conda install -c bioconda mash
# Single genome
mash sketch -o genome.msh genome.fasta
# Multiple genomes
mash sketch -o reference_db.msh genomes/*.fasta
# From reads (with coverage)
mash sketch -m 2 -r -o reads.msh reads.fastq.gz
# Pairwise distance
mash dist genome1.fasta genome2.fasta
# Query against database
mash dist reference_db.msh query.fasta > distances.tsv
# Screen for containment (metagenome)
mash screen reference_db.msh reads.fastq.gz > screen_results.tsv
| Distance | Interpretation |
|---|---|
| < 0.05 | Same species/strain |
| 0.05-0.15 | Same species |
| 0.15-0.25 | Same genus |
| > 0.25 | Different genus |
# All-vs-all distances
mash triangle genomes/*.fasta > distances.phylip
# Build tree
mash triangle -E genomes/*.fasta > distances.tsv
conda install -c bioconda sourmash
# Genome signature
sourmash sketch dna -p scaled=1000,k=31 genome.fasta -o genome.sig
# Multiple genomes
sourmash sketch dna -p scaled=1000,k=31 genomes/*.fasta -o genomes.sig
# Protein signatures
sourmash sketch protein -p scaled=100,k=10 proteins.faa -o proteins.sig
# Pairwise comparison
sourmash compare *.sig -o comparison.npy --csv comparison.csv
# Search against database
sourmash search query.sig database.sig --threshold 0.8
# Gather (metagenome decomposition)
sourmash gather metagenome.sig database.sig -o gather_results.csv
# Download taxonomy database
sourmash database download gtdb-rs214-k31.zip
# Classify
sourmash lca classify --db gtdb-rs214-k31.lca.json.gz --query query.sig
# Summarize metagenome
sourmash lca summarize --db gtdb-rs214-k31.lca.json.gz --query metagenome.sig
conda install -c bioconda fastani
# Single pair
fastANI -q query.fasta -r reference.fasta -o ani_result.txt
# Query vs multiple references
fastANI -q query.fasta --rl reference_list.txt -o ani_results.txt
# All-vs-all
fastANI --ql genome_list.txt --rl genome_list.txt -o all_vs_all.txt --matrix
| ANI | Interpretation |
|---|---|
| >99% | Same strain |
| 95-99% | Same species |
| <95% | Different species |
For strain-level analysis in metagenomes.
conda install -c bioconda instrain
# Map reads to reference
bowtie2 -x reference -1 reads_1.fq -2 reads_2.fq | \
samtools sort -o mapped.bam
# Profile with inStrain
inStrain profile mapped.bam reference.fasta -o instrain_output -p 8
# Profile multiple samples
for bam in sample*.bam; do
inStrain profile $bam reference.fasta -o ${bam%.bam}_IS -p 8
done
# Compare strain populations
inStrain compare -i sample*_IS -o comparison_IS -p 8
# SNV table
cat instrain_output/output/SNVs.tsv
# Gene-level info
cat instrain_output/output/gene_info.tsv
# Genome info
cat instrain_output/output/genome_info.tsv
Goal: Identify potential outbreak clusters by computing pairwise genomic distances across isolate genomes using multiple complementary methods.
Approach: Sketch genomes with MASH for fast distance estimation, compute ANI with fastANI for accurate species-level resolution, compare sourmash signatures for containment analysis, and cluster close matches to identify transmission pairs.
#!/bin/bash
set -euo pipefail
GENOMES_DIR=$1
OUTPUT_DIR=$2
mkdir -p $OUTPUT_DIR
echo "=== MASH sketching ==="
mash sketch -o $OUTPUT_DIR/genomes.msh $GENOMES_DIR/*.fasta
echo "=== MASH distances ==="
mash dist $OUTPUT_DIR/genomes.msh $OUTPUT_DIR/genomes.msh > $OUTPUT_DIR/mash_distances.tsv
echo "=== fastANI ==="
ls $GENOMES_DIR/*.fasta > $OUTPUT_DIR/genome_list.txt
fastANI --ql $OUTPUT_DIR/genome_list.txt \
--rl $OUTPUT_DIR/genome_list.txt \
-o $OUTPUT_DIR/fastani_results.txt \
--matrix
echo "=== sourmash signatures ==="
sourmash sketch dna -p scaled=1000,k=31 $GENOMES_DIR/*.fasta -o $OUTPUT_DIR/all.sig
sourmash compare $OUTPUT_DIR/all.sig -o $OUTPUT_DIR/sourmash.npy --csv $OUTPUT_DIR/sourmash.csv
echo "=== Identify clusters ==="
python3 << 'EOF'
import pandas as pd
import numpy as np
# Load MASH distances
mash = pd.read_csv('${OUTPUT_DIR}/mash_distances.tsv', sep='\t', header=None,
names=['ref', 'query', 'distance', 'pvalue', 'shared'])
# Filter for close matches (potential outbreak cluster)
close = mash[(mash['distance'] < 0.001) & (mash[] != mash[])]
()
(close[[, , ]])
EOF
import pandas as pd
import numpy as np
from scipy.cluster.hierarchy import linkage, fcluster
from scipy.spatial.distance import squareform
# Load MASH distances
mash = pd.read_csv('mash_distances.tsv', sep='\t', header=None,
names=['ref', 'query', 'dist', 'pval', 'shared'])
# Pivot to matrix
samples = sorted(set(mash['ref'].tolist()))
dist_matrix = mash.pivot(index='ref', columns='query', values='dist').fillna(0)
dist_matrix = dist_matrix.loc[samples, samples]
# Cluster
condensed = squareform(dist_matrix.values)
Z = linkage(condensed, method='average')
# Cut tree at species level (0.05)
clusters = fcluster(Z, t=0.05, criterion='distance')
cluster_df = pd.DataFrame({'sample': samples, 'cluster': clusters})
print(cluster_df.groupby('cluster').size())