In genome research, what is the problem in Mapping that may be caused by reads being too short?

Question

In the following scenario: You were given short sequence reads of plant RNA obtained from a next-generation sequencing machine (fragments of 20–30 nucleotides in length). You attempt to map them back to the genome, but a significant proportion of them do not align.

The question is: give some obvious explanations why the alignment of short sequences can fail, apart from possible contamination or technical difficulties during the preparation of the RNA.

I would answer it as because reads are short, and because of introns(since it is RNA)

Another scenario: There are some indications that the problematic sequences come from an uncharacterised plant RNA virus. What would you do next? What are the specific caveats with short sequence reads?

I got the above questions, I am a computer science student doing bioinformatics, any biologist could answer it will be well appreciated

Answer 1

Being a bioinformatician as well, I am not really what you asked for, but I work with plant genetics so I'll try and answer anyway.

What you are mapping is RNA. So, as you already figured out, splicing events will be a problem for end-to-end mapping of the reads. There are tools managing that, though, so let's assume you used one of them and still a lot of your reads do not map. To put in WYSIWIG's good point: Another event that can mess up with your alignments is RNA editing, although this is not too likely to cause a high proportion of reads to not align at all.

Some aligners might filter out "too short" queries, so make sure you are not using one of them.

Then, have you preprocessed your reads? If you have not, there might be adaptor sequences left. Or may reads have very poor quality, thus are also aligned with poor quality, thus might be counted as unaligned.

And then check what you are aligning to. Many published plant genomes are also of minor quality, including lots of unassigned bases. So, there could be large proportions of your reference genome that count in the genome length, but are only Ns and nothing will align there.

Last but not least, your thought about a virus could also be right. Depending on the experiment, there might be pathogenic RNA in your sample, so check against a suitable database.

If the problem is only that the reads are "too short" for whatever reason, try doing transcriptome assembly before comparing to your reference.

Answer 2

I don't think it's possible to answer part 1 without more information. Specifically, are you using a splice aware mapper, like Tophat? Are you using a predetermined gtf with putative exon coordinates? If so, is it for an extremely well studied plant, like Arabidopsis, or something brand new? Your answer isn't right, introns would align to the genome fine. The aligner doesn't know or care that those sequence shouldn't be in the sample, that won't affect mapping. Personally, I don't think there is a very good answer to the question. Short reads aren't more likely to fail than longer reads. They are more likely to map to the wrong place, but that's not failure to map.