使用rmats检测AS事件,用了越来越多后发现,当剪接事件很少时,会报告很多“难以验证”的AS事件(sashimiplot里面有对应的junction,但是看IGV很难看出区别)。
为了提高AS检测的准确性,现转向其他软件。
今天介绍一下SUPPA的原理和方法。
SUPPA计算差异剪接,主要分四步:
- 利用基因组注释建立可变剪接事件(ioe)或者转录本事件(ioi)
- 为每个样本的ioe或ioi定量PSI值
- 比较样本间的差异,计算dPSI
- 根据PSI值,对事件归类
SUPPA建立索引
python3.4 suppa.py generateEvents [options]
- -i gtf
- -o outprefix
- -f/–format ioe/ioi
- -e/–event-type (only used for local AS events)
- SE: Skipping exon (SE) events
- SS: Alternative 5’ (A5) and 3’ (A3) splice sites (it generates both)
- MX: Mutually Exclusive (MX) exons
- RI: Retained intron (RI)
- FL: Alternative first (AF) and last (AL) exons (it generates both)
- –pool-genes (Important). This option is important when creating ioe/ioi from annotations that are not loci-based, e.g.: RefSeq and UCSC genes. Unlike Ensembl or Gencode, which annotate gene loci, i.e. a set of transcripts will be uniquely be related to a gene at a locus, other annotations, like UCSC and Refseq dowloaded from UCSC, do not have this unequivocal link of transcripts to a genomic locus.
- 其他参数见SUPPA原页面
输出结果:运行完毕后,对于ioe,针对每个事件类型,会生成一个gtf格式文件和一个ioe文件,ioe文件用于后续计算。