strange bowtie output

[vebaev]

Hi all,
I have used bowtie with the following options as option -a result some reads mapping thousands times:
bowtie -f -m 200 -v 1 -p 3

As I know -m 200 will remove a tag which maps more than 200 times, but if maps 199 times they will be in the output?

As a result from -m 200 my output include only one maping (one alignment for one read? Even reads I know that map 5 times are showed as only 1 alignment?

How is that possible

[dreesbl]

@vebaev

As you found out, option -a is usually trouble! To answer your question, -m 200 means that NO alignments are reported for a read that has 200+ hits. For a read with 1, 2, or 199 alignments, the default behavior of Bowtie is to report a single alignment for the read. You can set the max number of alignments to report for each read with the -k option (default is -k 1).

Another thing to watch out for is that the single alignment reported is not necessarily the best one. Unless you use the --best option Bowtie will report the first valid alignment it encounters. The --best option sets it to find all valid alignments and report them in best to worst order (only the best one at -k 1), which is much more computationally intensive. Many of the "options" in Bowtie are shortcuts made necessary by limited computing resources and don't necessarily improve the quality of your results.

As an aside, why are you using read data in fasta format? What instrument platform is it from? Just curious why you don't have fastq or qseq. Bowtie takes base quality score into account and leaving it out will affect your alignment results. You could try BWA, which doesn't use base quality in its calculations for alignment.

[vebaev]

Thanks for the reply.
I have notised the details that you described.

I succseed outputing all alignments up to 200 for each read and discarding ones with 200+ by combining "-a -m 200"

I'm using fasta since I have already trimmed, filtered, cleaned and unified reads with their counts in fasta.