Hey Alex, Super helpful video thanks!! I was wonder what your thoughts on the pros/ cons of Portcullis vs STAR ability to find novel junctions? Also, would regtools work to find novel juctions with STAR *SJ.out.tab per sample?
Hey, Ileena. I am not sure about the technical comparisons to which one can find more or more accurate junctions but I'll give maybe my 2 cents. Portcullis is really nice because you can use any aligner that want which would allow for large gaps that is required for RNA-seq, like HISAT2 for example. Also, we could use STAR single pass not removing those alignments that contain junctions our annotation file lacks. The reason I chose Portcullis in this instance is because at work I was provided just BAM files and was still able to continue on with the novel junction discovery. Now, using STAR i think would be a easier or more streamlined approach if you started from the beginning and the whole analysis was yours. If you have the raw/trimmed reads, it's much easier to do a 2-pass alignment with STAR where the first time it uses the GTF as a sort of baseline, then the second time it adds high confidence junctions to the GTF and aligns again. If there's a new junction in that second alignment run that wasn't in the first STAR (I believe) discards that junction since it's likely false positive. As far as if RegTools would work that's a really good question. The output SJ.out.tab isn't *exactly* a bed file but the first few columns do contain the chr, start, end, and strand. However, the start and end are 'intronic' vs, I guess the more typical, 'exonic'. There might be ways to convert this to the BED12 that RegTools likes for annotation. Otherwise with RegTools there's actually and `extract` function too which can be used on the BAM file. Many different ways to accomplish the same task, and each has its pros and cons.
Dear Alex, thank you, it is very helpful. Do you have any experience with wSL2 or WSl1 with NGS analysis.
THANK YOU
thanks for sharing
Hey Alex, Super helpful video thanks!! I was wonder what your thoughts on the pros/ cons of Portcullis vs STAR ability to find novel junctions? Also, would regtools work to find novel juctions with STAR *SJ.out.tab per sample?
Hey, Ileena.
I am not sure about the technical comparisons to which one can find more or more accurate junctions but I'll give maybe my 2 cents.
Portcullis is really nice because you can use any aligner that want which would allow for large gaps that is required for RNA-seq, like HISAT2 for example. Also, we could use STAR single pass not removing those alignments that contain junctions our annotation file lacks. The reason I chose Portcullis in this instance is because at work I was provided just BAM files and was still able to continue on with the novel junction discovery.
Now, using STAR i think would be a easier or more streamlined approach if you started from the beginning and the whole analysis was yours. If you have the raw/trimmed reads, it's much easier to do a 2-pass alignment with STAR where the first time it uses the GTF as a sort of baseline, then the second time it adds high confidence junctions to the GTF and aligns again. If there's a new junction in that second alignment run that wasn't in the first STAR (I believe) discards that junction since it's likely false positive.
As far as if RegTools would work that's a really good question. The output SJ.out.tab isn't *exactly* a bed file but the first few columns do contain the chr, start, end, and strand. However, the start and end are 'intronic' vs, I guess the more typical, 'exonic'. There might be ways to convert this to the BED12 that RegTools likes for annotation. Otherwise with RegTools there's actually and `extract` function too which can be used on the BAM file.
Many different ways to accomplish the same task, and each has its pros and cons.
@@alexsoupir This is helpful to keep in mind as I get started. Thanks for your reply!