Rna-Seq alternative splicing analysis under limited conditions: possible?

Question

[user31888]

Being new to alternative splicing analysis jungle, what strategy/program(s) could you experts recommend knowing that, working on human, we have only 1 RNA-seq sample with the matched exome dataset (no replicates)? Is it at least possible?

I read so many reviews and tutorials, and heard about so many programs but they seem to be mainly able to handle differential analysis (replicates or different conditions). The only ones that seem to fit in my case would be splice-aware mappers (to only detect junctions) and maybe SpliceR, SplAdder, CASPER.

Also, I don't really understand the difference of output you can obtain from programs that reports splicing events only compared to the ones that produce isoforms only (although some can perform both - e.g. SOLAS, MISO, ALEXA-Seq which require multiple samples though). Do the former detect the type of splicing (SE, RI, ...) but don't give you the actual transcripts obtained from the event or you are still able to identify the transcripts?

[Devon Ryan]

What is your biological goal? Most people working on alternative splicing have a theory like, "If we do A, some splicing will change." So they measure things with two conditions and compare them. With just a single sample you're not really looking at alternative splicing, but just splicing. Perhaps you want to just look for novel isoforms or something like that.

[user31888]

Ideally, I would like to see if some transcripts from my RNA-seq data contain (pieces) of intronic sequences or missing exons (and which ones) compared to specific genes sequenced in my exome. Does it make sense?

[Sean Davis]

Exome sequencing does not tell you about gene or transcript structure, so I cannot see how comparing your RNA-seq to exome sequencing would answer any question relevant to alternative splicing. Could you elaborate?

[Amitm]

hi,

Maybe first process the RNA-seq data using any splice aware aligner you like, while giving a comprehensive GTF file (Gencode?). I use STAR and then StringTie but you could use any other mapper+ isoform quantifier combination. Then in the annotated SNVs from your exome, I would look for those annotated to be altering splice sites. These would be the lowest hanging fruits. For those SNVs I would look into the RNA-seq BAM to check if the splicing pattern is unusual.

Depending on your sample/ condition, you could use public databases (ENCODE, NCBI GEO etc.) for "normal" case splicing pattern.

Answer 1

[Sean Davis]

You have only one sample with both exome and transcriptome. I would suggest starting with STAR or tophat alignment to the appropriate reference genome and a set of transcripts (represented by a gtf file). Then, view the resulting BAM file in a browser such as IGV. From that exercise, you can very quickly determine if your hypotheses are even approximately true.