bio-workflows-metagenomics-pipeline

End-to-end shotgun metagenomics workflow from FASTQ to taxonomic and functional profiles, orchestrating controls/host depletion, Kraken2+Bracken classification, MetaPhlAn marker profiling, and HUMAnN functional profiling. Covers the controls-first ordering, why Kraken2 read counts are not abundances and MetaPhlAn cell fractions do not equal Bracken read fractions, and the consistent-pipeline framing. Use when profiling shotgun metagenomic samples end to end, or chaining classification, abundance, and function. For resistome see metagenomics/amr-detection; for strains see metagenomics/strain-tracking; for assembly see genome-assembly/metagenome-assembly.

跳到安装

仓库: GPTomics/bioSkills
最近来源活动: 2026年7月10日 19:25
检测到的 SKILL.md 语言: 英语
星标: 1,169
分支: 195

安装方式

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。

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/GPTomics/bioSkills --skill bio-workflows-metagenomics-pipeline

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。

检查来源文件

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。

文件资源管理器

3 个文件

SKILL.md

来源说明 · 只读预览

返回搜索返回安装

clawhub-installer

Installs 425 bioinformatics skills covering sequence analysis, RNA-seq, single-cell, variant calling, metagenomics, structural biology, and 56 more categories. Use when setting up bioinformatics capabilities or when a bioinformatics task requires specialized skills not yet installed.

2026-08-11

bio-seq-objects

Create and manipulate Seq, MutableSeq, and SeqRecord objects using Biopython. Use when creating sequences from strings, modifying sequence data in-place, building annotated records for file output, or debugging post-1.78 Bio.Alphabet and immutability errors.

2026-08-11

bio-transcription-translation

Transcribe DNA to RNA and translate to protein using Biopython, with NCBI codon-table selection, CDS validation, and six-frame ORF finding. Use when converting a CDS or ORF to its amino-acid sequence, selecting a non-standard (mitochondrial, bacterial, ciliate) genetic code, validating a coding sequence, or scanning all reading frames.

2026-08-11

查看这个仓库的全部 Skills

Metagenomics Pipeline

"Analyze my metagenomic samples from FASTQ to taxonomic and functional profiles" -> Orchestrate controls and host depletion, Kraken2/Bracken taxonomic classification, MetaPhlAn profiling, and HUMAnN3 functional analysis - reporting results relative to a consistent pipeline, never as a direct observation of the community.

Complete workflow from metagenomic FASTQ to taxonomic and functional profiles. This is a workflow skill: it owns the chaining decisions and hand-offs, not the internals of any one step.

The governing principle

A taxonomic/functional profile is a position in a choice-chain (extraction -> depletion -> depth -> classifier -> DB -> normalization), never a direct observation; the trustworthy result is decided at these seams.

The reference DB + version is THE inherited commitment. The Kraken2/GTDB (or standard/RefSeq/UHGG) DB chosen at classification fixes what is detectable — a zero means below-detection OR not-in-DB OR lost-in-extraction OR removed-by-depletion, almost never biological absence. Pin the DB build (version alone moves species/genus calls); match DB to habitat (UHGG for gut); report the CLASSIFIED FRACTION (a low fraction is the tell that the DB is wrong).

Host removal against T2T-CHM13 is a made-once commitment done BEFORE profiling. Prefer the complete T2T over gapped GRCh38 (which lets human reads masquerade as novel microbes); mask rDNA; discard both mates if either maps host. It is a privacy obligation (leaked human reads are identifiable), not just QC.

Controls-first is a design commitment, not a step added later. Extraction blanks + a whole-cell mock carried through the WHOLE workflow. NO low-biomass result (skin, BAL, CSF, blood, tissue) is interpretable without blanks + DNA-concentration + decontam; at near-zero biomass the signal IS the kitome (Salter 2014). A blank cannot be retrofitted.

Read-fraction is not cell-fraction, and tools/DBs are not comparable. Kraken2 read-fraction (genome-size/copy-number biased) and MetaPhlAn cell-fraction must never be merged into one table. Holding tool+DB constant within a study is the only way a comparison measures biology and not the tool.

Metagenomics Pipeline

Complete workflow from metagenomic FASTQ to taxonomic and functional profiles. This is a workflow skill: it owns the chaining decisions and hand-offs, not the internals of any one step.

The governing principle

#!/bin/bash set -e THREADS=8 KRAKEN_DB="kraken2_standard_db" HOST_INDEX="human_bt2_index" # MUST be built from T2T-CHM13 (CHM13v2, +HLA) per the made-once host-removal commitment, not a legacy GRCh38 index SAMPLES="sample1 sample2 sample3" OUTDIR="metagenomics_results" mkdir -p ${OUTDIR}/{trimmed,host_removed,kraken,bracken,metaphlan,humann,qc} # Step 1: QC echo "=== QC ===" for sample in $SAMPLES; do fastp -i ${sample}_R1.fastq.gz -I ${sample}_R2.fastq.gz \ -o ${OUTDIR}/trimmed/${sample}_R1.fq.gz \ -O ${OUTDIR}/trimmed/${sample}_R2.fq.gz \ --length_required 50 \ --html ${OUTDIR}/qc/${sample}_fastp.html -w ${THREADS} done # Host removal echo "=== Host Removal ===" for sample in $SAMPLES; do # -f 12 (read unmapped AND mate unmapped) keeps only pairs where NEITHER mate hit the host. # --un-conc-gz would instead keep every non-CONCORDANT pair, retaining pairs whose mate mapped # human -- a privacy leak, not just a QC lapse. -F 256 drops secondary alignments. bowtie2 -p ${THREADS} -x ${HOST_INDEX} --very-sensitive \ -1 ${OUTDIR}/trimmed/${sample}_R1.fq.gz \ -2 ${OUTDIR}/trimmed/${sample}_R2.fq.gz \ 2> ${OUTDIR}/qc/${sample}_host.log \ | samtools view -b -f 12 -F 256 - \ | samtools sort -n -@ ${THREADS} - \ | samtools fastq -1 ${OUTDIR}/host_removed/${sample}_R1.fq.gz \ -2 ${OUTDIR}/host_removed/${sample}_R2.fq.gz \ -0 /dev/null -s /dev/null -n - done # Step 2: Kraken2 echo "=== Kraken2 ===" for sample in $SAMPLES; do kraken2 --db ${KRAKEN_DB} --threads ${THREADS} --paired \ --confidence 0.1 --minimum-hit-groups 2 \ --report ${OUTDIR}/kraken/${sample}.report \ --output ${OUTDIR}/kraken/${sample}.output \ ${OUTDIR}/host_removed/${sample}_R1.fq.gz \ ${OUTDIR}/host_removed/${sample}_R2.fq.gz done # Bracken echo "=== Bracken ===" for sample in $SAMPLES; do bracken -d ${KRAKEN_DB} \ -i ${OUTDIR}/kraken/${sample}.report \ -o ${OUTDIR}/bracken/${sample}.species.txt \ -r 150 -l S -t 10 done echo "=== Pipeline Complete ===" echo "Kraken reports: ${OUTDIR}/kraken/" echo "Bracken abundances: ${OUTDIR}/bracken/"

name	bio-workflows-metagenomics-pipeline
description	End-to-end shotgun metagenomics workflow from FASTQ to taxonomic and functional profiles, orchestrating controls/host depletion, Kraken2+Bracken classification, MetaPhlAn marker profiling, and HUMAnN functional profiling. Covers the controls-first ordering, why Kraken2 read counts are not abundances and MetaPhlAn cell fractions do not equal Bracken read fractions, and the consistent-pipeline framing. Use when profiling shotgun metagenomic samples end to end, or chaining classification, abundance, and function. For resistome see metagenomics/amr-detection; for strains see metagenomics/strain-tracking; for assembly see genome-assembly/metagenome-assembly.
tool_type	cli
primary_tool	Kraken2
workflow	true
depends_on	["read-qc/fastp-workflow","metagenomics/contamination-controls","metagenomics/kraken-classification","metagenomics/metaphlan-profiling","metagenomics/abundance-estimation","metagenomics/functional-profiling","metagenomics/metagenome-visualization"]
qc_checkpoints	[{"after_qc":"Q30 >80%, host reads removed"},{"after_classification":"Classification rate >60%, known taxa dominant"},{"after_functional":"Pathway coverage reasonable, unmapped <50%"}]

name	bio-workflows-metagenomics-pipeline
description	End-to-end shotgun metagenomics workflow from FASTQ to taxonomic and functional profiles, orchestrating controls/host depletion, Kraken2+Bracken classification, MetaPhlAn marker profiling, and HUMAnN functional profiling. Covers the controls-first ordering, why Kraken2 read counts are not abundances and MetaPhlAn cell fractions do not equal Bracken read fractions, and the consistent-pipeline framing. Use when profiling shotgun metagenomic samples end to end, or chaining classification, abundance, and function. For resistome see metagenomics/amr-detection; for strains see metagenomics/strain-tracking; for assembly see genome-assembly/metagenome-assembly.
tool_type	cli
primary_tool	Kraken2
workflow	true
depends_on	["read-qc/fastp-workflow","metagenomics/contamination-controls","metagenomics/kraken-classification","metagenomics/metaphlan-profiling","metagenomics/abundance-estimation","metagenomics/functional-profiling","metagenomics/metagenome-visualization"]
qc_checkpoints	[{"after_qc":"Q30 >80%, host reads removed"},{"after_classification":"Classification rate >60%, known taxa dominant"},{"after_functional":"Pathway coverage reasonable, unmapped <50%"}]

Commitment	Consequence inherited downstream
Reference DB + version + habitat match	What is detectable; a zero is below-detection/not-in-DB, not absence; report classified fraction
Host-removal reference (T2T-CHM13)	Human reads masquerading as microbes; a privacy obligation; removed before profiling
Controls (blanks + mock) carried through	Whether any low-biomass result is interpretable; the signal IS the kitome without them
Extraction method held constant	Extraction bias outweighs much biological signal (Costea 2017); interacts with read-vs-assembly choice

Symptom	Cause	Fix
A "novel community" that is the kitome	Negative controls skipped / contamination bleed	Blanks + mock through the full workflow; decontam; skepticism toward canonical kitome genera
Same organism appears twice under different names	Merged GTDB-Tk-labelled and NCBI-labelled tables (Firmicutes vs Bacillota)	State which taxonomy each table uses; never name-merge without a crosswalk
Cross-study comparison is really tool differences	Compared across tools/DBs	Hold tool + DB constant within a study; benchmark on a mock with OPAL
A zero read as biological absence	Confused detection-limit with biology	Name which link (depth/DB/extraction/depletion) is responsible before any biological reading; Nonpareil depth check
Read-fraction and cell-fraction merged	Kraken2 % joined to MetaPhlAn %	Keep separate tables; they use different absence semantics
Low classification rate	Database mismatch / novel organisms	Match DB to habitat; report classified fraction; a low fraction = wrong/incomplete DB
High host reads	Incomplete host removal	Use the complete T2T host reference; mask rDNA

bio-workflows-metagenomics-pipeline

安装方式

检查来源文件

Version Compatibility

Metagenomics Pipeline

The governing principle

Version Compatibility

Metagenomics Pipeline

The governing principle

Made-once commitments

Workflow Overview

Primary Path: Kraken2 + Bracken + HUMAnN

Step 0: Quality Control, Host Removal, and Controls

Step 1A: Kraken2 Classification

Step 1B: Bracken Abundance Estimation

Step 1C: Alternative - MetaPhlAn Profiling

Step 2: Functional Profiling with HUMAnN

Visualization

Parameter Recommendations

Common Errors

References

Complete Pipeline Script

Related Skills

Step	Parameter	Value
fastp	--length_required	50 (metagenomic reads)
Kraken2	--confidence	0.1-0.4 (default 0.0 over-classifies; see metagenomics/kraken-classification)
Kraken2	--minimum-hit-groups	2 (cut single-region false positives)
Bracken	-r	Read length (e.g., 150; must match the DB build)
Bracken	-l	S (species) or G (genus)
Bracken	-t	10 (min reads threshold)
MetaPhlAn	--min_cu_len	2000 (default)
HUMAnN	--threads	8+

bio-workflows-metagenomics-pipeline

来源信息

安装方式

检查来源文件

同仓库更多 Skills

同仓库更多 Skills

Version Compatibility

Metagenomics Pipeline

The governing principle

Version Compatibility

Metagenomics Pipeline

The governing principle

Made-once commitments

Workflow Overview

Primary Path: Kraken2 + Bracken + HUMAnN

Step 0: Quality Control, Host Removal, and Controls

Step 1A: Kraken2 Classification

Step 1B: Bracken Abundance Estimation

Step 1C: Alternative - MetaPhlAn Profiling

Step 2: Functional Profiling with HUMAnN

Visualization

Parameter Recommendations

Common Errors

References

Complete Pipeline Script

Related Skills