Workflow for identification of new splicing variants from RNA-seq data?

Question

I'd like to perform a search for possibly unidentified splicing variants of a specific protein in A. Thaliana. I do not have my own RNA-seq data. It has been suggested to me to use this workflow:

1. Download RNA-seq data from SRA with fastq-dump.
2. Construct an index of reference sequences with bowtie2-build.
3. Map the reads from RNA-seq data to the reference sequences with bowtie2.
4. Remove non-specific mappings with samtools view -q 10. Sort the mappings with samtools sort.
5. Browse the aligned reads with tablet.

I have been given example data of another organism (not A. Thaliana) and have successfully completed these steps. However, I don't know how to apply that in my research. Specifically, I'd like to know:

1. Is there other/better way to do do this?
2. Can I narrow the search/alignment to a specifc gene?
3. Where can I download A. Thaliana's reference genome in FASTA format?
4. Which SRA runs should I choose for my analysis?
5. How to interpret the results to get the information on the different splicing variants?

Linking a similar question I asked before: How to find an unidentified splicing variant of a protein?

Answer 1

Is there other/better way to do do this?

There is almost always a better way for pretty much anything but in this case I would stick with what your colleagues are using/suggesting. Once you have a worked-out case, you can always improve if needed. Other than that what is suggested to you seems reasonable to me (I have done sth similar for a gene family of some 100 genes using a workflow similar to what was suggested to you)

Can I narrow the search/alignment to a specifc gene?

Technically yes, instead of the whole genome, you can use a subset of the genome. This will present some problems, for example since the aligner will not be able to find native "targets" for most of the reads, some reads will map to your "new reference" (genome subset) in a "False-positive" manner. You will need to play with the settings. I would advise against this.

Where can I download A. Thaliana's reference genome in FASTA format?

Genbank, EMBL, ... There might be a specific repo for your organism like there is Flybase for Drosophila species and Xenbase for Xenopus species.

Which SRA runs should I choose for my analysis?

RNA-seq experiments conducted on your species of interest and ideally on a tissue where your genes of interest are highly expressed. If there is none, you could use a similar organism but this would come with some problems. The most ideal would be RNA-seq with long-read sequencing (like Pacbio).

How to interpret the results to get the information on the different splicing variants? When you visualize the aligned reads and compare these with existing gene annotation model, it will immediately be obvious if there are novel isoforms, could be due to "new" or not-yet-known exons/splice signals.