Detecting a small set of given Intron-Exon junctions in a RNA-Seq dataset

Question

[Pierre Lindenbaum]

Hi all, I've been given a set of RNA-Seq fastqs (4 samples) and a collaborator would like to know how many defined intron-exon junction in his favorite gene are detected in those reads.

My idea was to clean the reads with cutadapt, align with Bowtie and create a custom program to loop over the cigar string and detecting the junctions.

or is there an existing tool for this ? Bonus if the tool can tell me wether two junctions belong to the same transcript.

Answer 1

[igor]

If you use the STAR aligner, it will output SJ.out.tab with high confidence collapsed splice junctions in tab-delimited format.

[Pierre Lindenbaum]

that was simple and it workded fine, thanks

Answer 2

[Malachi Griffith]

You can use the tool RegTools to get this information from a BAM file. So you would align your reads to the reference genome first using TopHat2, STAR, HISAT2, etc. Then use regtools to both extract and annotate the exon-exon junctions (i.e. exon-intron ... intron-exon). The

Code is here: https://github.com/griffithlab/regtools

Documentation is here: https://regtools.readthedocs.io/en/latest/

You would start with regtools junctions extract followed by regtools junctions annotate

Usage looks like this:

regtools junctions extract [options] indexed_alignments.bam
regtools junctions annotate [options] junctions.bed ref.fa annotations.gtf

The annotate result will among other things, tell you which junctions correspond to which transcripts, what novel junctions might be observed in your RNA-seq data, how these relate to known transcripts, etc. The annotations.gtf file could be a list of known transcript annotations. For example, we often use the GTF files that you can down directly from Ensembl.

Answer 3

[Khader Shameer]

I have used MapSplice, better to use their recent version MapSplice 2

MapSplice: Accurate mapping of RNA-seq reads for splice junction discovery