Flow Map流程地图
The TAFFISH RNA-seq suite is a family of small, versioned, auditable flow apps. The reference route starts from a genome and annotation; the de novo route starts from reads and builds the transcript feature space first. Both routes keep command provenance, logs, output contracts, and static reports explicit.
TAFFISH RNA-seq 套件由一组小而清楚、可版本化、可审计的流程应用组成。有参路线从基因组和注释开始；无参路线从 reads 开始，先构建转录本特征空间。两条路线都显式保留命令溯源、日志、输出契约和静态报告。
In the current release set, core subflows are documented as 0.2.0-r1, rnaseq-standard-flow 0.3.0-r1 bridges selected internal tool-level @step-name: advanced slots, and rnaseq-report-flow 0.3.0-r2 produces standalone reports with embedded plots, tables, and real QC/report HTML payloads.
当前版本组合中，核心子流程为 0.2.0-r1，rnaseq-standard-flow 0.3.0-r1 会桥接部分内部 tool-level @step-name: 高级参数槽，rnaseq-report-flow 0.3.0-r2 则生成内嵌图、表和真实 QC/报告 HTML payload 的 standalone 报告。
Reference mode remains the default for projects with a reliable genome and annotation. The optional alignment lane adds genome-aware evidence. De novo mode is an explicit no-reference route that replaces the reference index and expression segment with assembly, de novo quantification, and annotation.
有参模式仍是拥有可靠基因组和注释项目的默认路线。可选比对分支补充基因组位置证据。无参模式是显式 no-reference 路线，用组装、无参定量和注释替代有参 index 与 expression 段。
--route both; DE can switch to featureCounts with --de-source featurecounts
通过 --route both 启用；差异分析可用 --de-source featurecounts 切换到 featureCounts
--mode denovo; no genome or annotation is required, but annotation resources must be supplied for functional interpretation
通过 --mode denovo 启用；不需要基因组或注释，但功能解释需要用户提供注释资源
rnaseq-standard-flow is designed for users who want to start from local FASTQ files and receive a coherent analysis directory plus a bilingual HTML report. In reference mode, the inputs are FASTQ, genome FASTA, annotation, metadata, and optional gene sets. In de novo mode, the inputs are FASTQ, metadata, a protein database for homology annotation, and an optional protein-to-GO map for enrichment.
rnaseq-standard-flow 面向希望从本地 FASTQ 开始，并得到一个完整分析目录和双语 HTML 报告的用户。有参模式输入 FASTQ、基因组 FASTA、注释、元数据和可选基因集；无参模式输入 FASTQ、元数据、用于同源注释的蛋白数据库，以及用于富集分析的可选蛋白到 GO 映射表。
The default behavior remains reference Salmon-first. De novo mode is never entered silently just because a genome was omitted; users must select --mode denovo. This protects routine reference analyses from accidental mode switches and makes no-reference interpretation boundaries visible in the final report.
默认行为仍然是有参 Salmon-first。流程不会因为用户漏传 genome 就悄悄切到无参；必须显式选择 --mode denovo。这样可以避免常规有参分析被意外切换，也能让无参解释边界在最终报告中清楚呈现。
Advanced tuning is available without turning the route into a black box. The umbrella flow exposes selected internal subflow/tool slots with names such as @denovo-assembly-trinity-assembly-step:; these slots are empty by default and are used only when an expert explicitly passes extra native tool options.
高级调参不会把流程变成黑盒。总流程会公开部分内部子流程/工具槽位，例如 @denovo-assembly-trinity-assembly-step:；这些槽位默认为空，只有高级用户显式传入底层工具原生参数时才生效。
For the most complete command-line manuals, including parameter-by-parameter reference and full reference/de novo examples, use the executable flow documentation in the rnaseq-standard-flow GitHub docs.
如果要正式分析自己的数据，最完整的命令行说明、逐参数解释和有参/无参完整示例，请阅读 rnaseq-standard-flow GitHub docs。
Builds the reusable reference contract from a genome and annotation: standardized annotation, transcript FASTA, tx2gene.tsv, Salmon/Kallisto indexes, and optional HISAT2 genome index.
从基因组和注释构建可复用参考契约：标准化注释、转录本 FASTA、tx2gene.tsv、Salmon/Kallisto 索引，以及可选 HISAT2 基因组索引。
Quantifies reads against a reference transcriptome with Salmon, imports transcript evidence to gene-level matrices through tximport, and summarizes read/quantification QC.
使用 Salmon 将 reads 定量到有参转录组，通过 tximport 汇总到基因层面矩阵，并汇总 reads 与定量 QC。
Starts the no-reference route. It performs read QC/trimming, Trinity-first transcriptome assembly, transcript filtering, assembly statistics, optional offline BUSCO, and read-support summaries.
无参路线第一步。它完成 reads QC/修剪、Trinity-first 转录组组装、转录本过滤、组装统计、可选离线 BUSCO 和 reads 支持度摘要。
Builds a Salmon index from assembled transcripts, quantifies each sample, and emits transcript-level count/TPM matrices. Gene or cluster matrices are produced only when an explicit mapping is supplied.
从组装转录本构建 Salmon 索引，对每个样本定量，并输出转录本层面 count/TPM 矩阵。只有显式提供映射时，才生成 gene 或 cluster 层面矩阵。
Predicts ORFs with TransDecoder, searches predicted proteins against a user-provided protein database with DIAMOND, writes annotation tables, and can derive transcript-space GMT/background files from a GO map.
使用 TransDecoder 预测 ORF，通过 DIAMOND 将预测蛋白搜索到用户提供的蛋白数据库，输出注释表，并可根据 GO 映射派生转录本空间的 GMT/background。
Runs DESeq2 from a count matrix and metadata. In reference mode this is usually gene-level DE; in de novo mode it may be transcript-level unless a stable mapping was supplied.
基于计数矩阵和样本元数据运行 DESeq2。有参模式通常是基因层面差异；无参模式如果没有稳定映射，则多为转录本层面差异。
Interprets DE at gene-set or transcript-set level through offline GMT-based ORA and GSEA. The input universe must match the feature ID space used for DE.
通过离线 GMT 进行 ORA 和 GSEA，把差异结果提升到 gene set 或 transcript set 层面解释。输入背景必须匹配 DE 使用的特征 ID 空间。
Maps reads back to a reference genome with HISAT2 and produces sorted BAM files. It is useful for genome browser inspection, alignment evidence, featureCounts, and BAM QC.
使用 HISAT2 将 reads 比对回参考基因组并生成排序 BAM。适合基因组浏览器查看、比对证据、featureCounts 和 BAM 质控。
Converts aligned BAM evidence into a featureCounts gene-level matrix. It supports an alignment-derived DE source and comparison against Salmon/tximport counts.
把比对 BAM 转换成 featureCounts 基因计数矩阵，可作为基于比对的 DE 来源，也可与 Salmon/tximport 计数交叉比较。
Evaluates BAM-level evidence with SAMtools, RSeQC, Qualimap, and MultiQC. It explains whether the alignment branch is technically trustworthy.
结合 SAMtools、RSeQC、Qualimap 和 MultiQC 评估 BAM 层面证据，说明比对分支在技术上是否可信。
Collects upstream outputs into a bilingual standalone report and interpretation guide. It embeds plots, key tables, and real QC HTML payloads so the main report can be shared as a single HTML file.
把上游输出收集成双语 standalone 报告和解读指南。它内嵌图、关键表格和真实 QC HTML payload，使主报告可以作为单个 HTML 文件分发。
The public end-to-end entrypoint. It composes the selected subflows, exposes advanced internal-tool step passthrough, keeps output blocks, collects split PNG/PDF plots, and generates the final standalone report.
面向用户的一站式入口。它组合被选择的子流程，开放高级内部工具 step 参数桥接，保留各自输出块，收集拆分的 PNG/PDF 图，并生成最终 standalone 报告。
Use --route both only in reference mode. De novo mode does not create genome-aligned BAM, featureCounts matrices, or reference-based BAM QC.
只有有参模式才使用 --route both。无参模式不会生成基因组比对 BAM、featureCounts 矩阵或基于参考的 BAM 质控。