作用：比较已发表的 Gifu (v1.2) 与 MG20 (gnm3) 组装的 gap 数目与长度，并写出统计表与对比图。
入口：gap_analysis.py
输入：与 gap_analysis/ 同级的 Gifu/Gifu_gap.gff、Gifu/Lotusjaponicus_Gifu_v1.2_genome.fa.fai、MG20/Lotus_MG20_gap.gff、MG20/Lotus_japonicus.fasta.fai（GFF3 中 type 为 gap，1-based 闭区间）；染色体长度取自 .fai。
输出：results/（{Gifu,MG20}_gap_summary_per_chromosome.csv、{key}_gap_positions.csv、{key}_gap_size_distribution.csv、gap_genome_comparison.csv）与 figures/（{key}_gap_per_chromosome.png、{key}_gap_size_distribution.png、{key}_gap_map.png、comparison_genome_level.png、comparison_per_chromosome.png、comparison_size_ecdf.png）。
运行：python gap_analysis.py（无绝对路径：数据、results/、figures/ 均由脚本位置推导，Gifu/、MG20/ 必须与 gap_analysis/ 同级）。
工具：Python 3 + pandas、numpy、matplotlib（Agg）；字体优先 WenQuanYi Micro Hei，回退 DejaVu Sans；染色体排序正则 (?:chr|Lj)(\d+|[A-Za-z]+)$。
English
**Purpose**: Compare gap number and length between the published Gifu (v1.2) and MG20 (gnm3) assemblies and write the statistics tables and comparison plots.
**Entry point**: `gap_analysis.py`
**Inputs**: in the directory next to `gap_analysis/`: `Gifu/Gifu_gap.gff`, `Gifu/Lotusjaponicus_Gifu_v1.2_genome.fa.fai`, `MG20/Lotus_MG20_gap.gff`, `MG20/Lotus_japonicus.fasta.fai` (the GFF3 records with type `gap`, 1-based closed intervals); chromosome lengths come from the `.fai`.
**Outputs**: `results/` (`{Gifu,MG20}_gap_summary_per_chromosome.csv`, `{key}_gap_positions.csv`, `{key}_gap_size_distribution.csv`, `gap_genome_comparison.csv`) and `figures/` (`{key}_gap_per_chromosome.png`, `{key}_gap_size_distribution.png`, `{key}_gap_map.png`, `comparison_genome_level.png`, `comparison_per_chromosome.png`, `comparison_size_ecdf.png`).
**Run**: `python gap_analysis.py` (no absolute paths: the data, `results/` and `figures/` are derived from the script location, so `Gifu/` and `MG20/` must sit next to `gap_analysis/`).
**Tools**: Python 3 with pandas, numpy and matplotlib (`Agg`); font preference `WenQuanYi Micro Hei`, falling back to `DejaVu Sans`; chromosome order regex `(?:chr|Lj)(\d+|[A-Za-z]+)$`.