Beginner Questions

[tedwong]

I'm a beginner to the field. I'm trying to align my reads to a reference genome. However, I'm not quite sure I understand what tophat is trying to do. In fact, I want to make sure my understanding to RNA sequencing is correct.

Let's say,

"tophat -p 8 -G genes.gtf -o C1_R1_thout genome C1_R1_1.fq C1_R1_2.fq"

From my understanding, the *.fq files are raw sequence reads generated by experiments. Those raw reads are aligned to sequences defined in a reference genome. The reference genome in this example is "genome". genes.gtf lists some known transcripts.

The point of doing it is to analyse unknown transcripts. Say, if some of my reads can't be aligned to the reference genome, we know it might mean that there's a new mutation. Am I correct?

Another question, why do we need to provide an annotation file (genes.gtf) to tophat? Tophat already has the information from the reference genome to align with?

Thanks,

[dpryan]

The point of doing it is to analyse unknown transcripts. Say, if some of my reads can't be aligned to the reference genome, we know it might mean that there's a new mutation. Am I correct?

Tophat helps serve a few different and sometimes overlapping roles. Firstly, it will align the reads to known (annotated) transcripts. Secondly, it will align reads directly to the genome. Thirdly, it will attempt to align reads in such a way that finding novel (unannotated) transcripts is possible. For actually finding these, tools like cufflinks are useful.

There are also many reasons a given read won't align to the the genome/transcriptome. Firstly, it could be from a contaminant (bacteria, etc.). This is extremely common. Secondly, it could be from a portion of the genome that doesn't exist in the reference sequence (e.g., the 45S rRNA sequence isn't in the mouse genome and there are a number of sequence gaps in most (all?) mammalian genomes). Thirdly, there may be variants in the sample that was sequenced that aren't present in the reference sequence (there's almost always some of this). Finally, this can be caused by sequencing errors.

There are undoubtedly other reasons in addition to those I listed above!

Another question, why do we need to provide an annotation file (genes.gtf) to tophat? Tophat already has the information from the reference genome to align with?

Providing tophat with a reference annotation allows it to more easily align reads to the known transcriptome. Most of you reads will align either to that or to the genome, so this decreases the number of reads that need to be used to find novel splice sites. Finding these is somewhat computationally demanding and not without its issues.