Originally posted by sshen
View Post
These options can be difficult to fully understand at first. If you haven't yet, carefully read the Bowtie manual (http://bowtie-bio.sourceforge.net/manual.shtml) paying special attention to all of the examples under the Reporting Modes section.
There are two primary groups of options in Bowtie which impact the alignments which are reported. The first group (-n, -v, -e, -l, -I, -X, --fr, --rf, --ff) relate to defining what is or is not a valid alignment. The second group, and the one you are asking about (-k, -a, -m, -M, --best, --strata) affect which and how many of these valid alignments are reported in the output.
-k simply limits the maximum number of alignments reported. The argument to -k is any positive integer (1..infinity); -a is simply a special case of -k meaning "all". Bowtie must use one or the other of these options, but if you don't explicitly include it on the command line Bowtie will use its default of k=1.
If by the setting of -k (>1) or -a the valid, reportable alignments fall into more than one stratum (please read the manual for the definition of stratum), the adding the --strata option will further limit the reported alignments, only including those with the same number of mismatches in the seed (if -n mode was used) or in the whole alignments (if -v was used).
Note that if you use --strata you MUST also use --best.
-m is not required to use --strata but the two options can combine to further limit the reported alignments.
It should also be pointed out that in the example command line you provided --strata is meaningless. You have already limited the output to one best alignment (-k 1 --best). --strata will only come into play if you are first allowing more than one reportable alignment (-k > 1 or -a).
Leave a comment: