A feature that doesn’t exist on most grep-like tools is the ability to count the number of matches found per-pattern. This is actually quite a useful feature for bioinformatics and shows up often when exploring data.
Let’s pretend that we have a collection of patterns - perhaps these are sections of a transcript, or a list of cell-barcodes, or perhaps sgRNA protospacers - and we want to count how many times each of them appear in our dataset.
Let’s take a look at these patterns:
bqtools grep will accept a text file containing a list of patterns to search for - so let’s count how many times each of these patterns appear in our dataset:
Note that we use the -P/--pattern-count flag to enter the pattern-counting mode of bqtools grep.
This will give us a table with 3 columns that reflect the pattern, the number of sequence matches, and the fraction of all records that match that pattern.
If you have a large number of patterns and they are not regular expressions, try using the -x/--fixed flag which makes use of an optimized algorithm to count the number of matches.