作用：bulk RNA-seq 上游处理——fastp 质控、HISAT2 比对到 Gifu T2T 参考、StringTie 逐样本定量，并把各样本 TPM 列合并为一张矩阵。
入口：fastp.sh、align.sh
输入：align.conf（样本名、R1 FASTQ、R2 FASTQ 三列）及其指向的 raw/<sample>-R1.fq.gz、-R2.fq.gz；建索引用 Gifu_v1.0.fasta，StringTie 用 Lotus_GifuT2T.gtf。
输出：clean/<sample>-R{1,2}.fastq.gz、report/<sample>.json|.html、排序并建索引的 <sample>.bam(.bai)、02.stringtie/<sample>/<sample>.gtf|.tsv、02.tpm/<sample>.tpm 与合并矩阵 02.tpm/all_samples.tpm.tsv。
运行：bash fastp.sh、bash align.sh（fastp.sh 须在本目录内运行；align.sh 的 $index 从未赋值、-x 拿不到索引路径，且须以 01.align 为当前目录，运行前需修正脚本）。
工具：fastp（-w 88，JSON/HTML 报告）、HISAT2（hisat2-build -p 64；-p 64 --new-summary）、samtools（view -H、sort -@ 64、index）、StringTie（-p 64 -G -e -B -o -A）、Awk 与 sort/join 合并 TPM。
English
**Purpose**: Upstream bulk RNA-seq processing — fastp read QC, HISAT2 alignment to the Gifu T2T reference and StringTie per-sample quantification, joined into one TPM matrix.
**Entry point**: `fastp.sh`, `align.sh`
**Inputs**: `align.conf` (sample name, R1 FASTQ, R2 FASTQ) and the `raw/-R1.fq.gz` / `-R2.fq.gz` files it points to; `Gifu_v1.0.fasta` for the index and `Lotus_GifuT2T.gtf` for StringTie.
**Outputs**: `clean/-R{1,2}.fastq.gz`, `report/.json|.html`, the sorted and indexed `.bam(.bai)`, `02.stringtie//.gtf|.tsv`, `02.tpm/.tpm` and the joined matrix `02.tpm/all_samples.tpm.tsv`.
**Run**: `bash fastp.sh`, `bash align.sh` (`fastp.sh` must run in this directory; in `align.sh` `$index` is never set, so `-x` receives no index path, and `01.align` must be the current directory — fix the script before running).
**Tools**: fastp (`-w 88`, JSON/HTML reports), HISAT2 (`hisat2-build -p 64`; `-p 64 --new-summary`), samtools (`view -H`, `sort -@ 64`, `index`), StringTie (`-p 64 -G -e -B -o -A`), Awk with `sort`/`join` for the TPM merge.