Author: John S. Hawkins [[email protected]]
- bowtie
IMPORTANT NOTE: bowtie2 is not just "the new version", it's actually functionally different. This code specifically requires that you have installed bowtie. bowtie2's differences preclude its use for this purpose. It's fine if you have both installed, provided that 'bowtie' resolves to the non-bowtie2 version in your enviroment.
- bowtie-build (should come with bowtie)
- the Biopython library suite for python (can be installed with pip)
- the pysam library for python (can be installed with pip)
Primarily you will use the script buid_sgrna_library.py. See
build_sgrna_library.py -h
for usage information.
The normal usage case is to call the script with a genbank file like so:
./build_sgrna_library.py --input_genbank_genome_name testdata/U00096.3_full_sequence.gb
Which will generate an adjacent file called
testdata/U00096.3_full_sequence.targets.all.tsv
which specified all of the targets for the provided genome (the test genome, in this case), annotated with the locus_tag, and scored for specificity (the final column)
For bacteria we suggest using guides that
- have a small, positive offset
- are on the antisense strand ('anti' in the 'transdir' column)
- have a SPECIFICITY score of 39
If a guide meeting these criteria is not available, lower specificity can be used, but you should check for near-matches elsewhere in the genome to see if they are likely to cause issues. Guides on the 'sense' strand are not recommended. They generally have a greatly reduced, and hard to predict, level of effect. If reduced effect is desired, we suggest the use of http://www.github.com/traeki/mismatch_crispri to achieve more reliable outcomes.