Bowtie - multiple alignments and seed length vs read length in miRNA mapping

Question

[MH85]

Hi everyone,

I have miRNA reads with a mean length of 21bp. I want to map these reads onto human miRNA reference sequences (miRBase) that have a mean length of 22bp (ranging from 16bp - 28bp).

The current command is

bowtie --seedmms 1 -p 1 -k 1  --seedlen 25 --best

Q1. With -k 1 and --best, what happens if one read maps onto two references with exactly the same mapping values?

Q2. What happens when seed is longer than read? Should not be seed set always shorter than the length of the read?

Thanks a lot in advance

Milos

Answer 1

[Trivas]

1. Most miRNA references I've seen "collapse" identical miRNA sequences and identify it as say Let7a/7c/7d. This removes any issues with mapping to two separate references identically.
2. If seed is longer than the read, you won't get any mapped reads. Try it out - miRNA libraries are small enough that a single bowtie run takes seconds-minutes.