背景

近期做的RNA-Seq数据上游，hisat2+stringtie流程，使用的stringtie官方的prepDE.py脚本定量。

prepDE.py -i sample_lst.txt

命令报错：

Error: could not locate transcript ENST00000624361 entry for sample Sample10

Traceback (most recent call last):

  File "/share/home/xiehs/tools/stringtie-2.1.4.Linux_x86_64/prepDE.py", line 281, in <module>

    geneDict.setdefault(geneIDs[i],{}) #gene_id

KeyError: 'ENST00000624361'

解决方案

经过一系列的google探索，得出了这是因为前面使用的stringtie版本不对，下面贴出官方的回答：

Thank you so much @Gin-Wang I took a look at the data you uploaded and I can confirm now that the problem is real and correctly reported by users in this thread (and in #234), stringtie 2.0 with the -e option outputs a few STRG entries which should NOT be there, as the output should only contain the MSTRG entries that were given in the -G file.

We will fix this -e problem soon and report back here, until then please use stringtie v1.3.6 for quantification via prepDE (for short reads there shouldn't be a significant difference).

意思就是选用1.3.6的版本，他们会尽快修复。

那么如何快速的切换版本呢？自然是使用conda管理。

conda activate test

conda install stringtie=1.3.6-0

使用我们就用绝对路径写命令即可。

/share/Software/miniconda3/envs/test/bin/stringtie

跑完gtf文件，用脚本定量，果然没有报错了。

特此分享给大家。