Wrong number of predicted mismatches

[DaniPaulo]

Hi,

I'm trying to use Cas-Offinder (conda version or v.2.4) with Cas-designer (v.1.2 standalone) and Cas-Offinder-bulge (v.1.2 beta), and – although the pipeline runs – I'm getting some wrong results regarding the number of mismatches.

command:
python3 cas-designer file.config

where 'file.config' looks like this:

/path-to/genome.fasta
/path-to/target.fasta
20
NGG
NGG
5
0
0

I noticed the mismatch count is wrong by inspecting the output 'file-offtargets.txt', for example:

#Bulge type	crRNA	DNA	Chromosome	Position	Direction	Mismatches	Bulge Size
X	TTCACCCGGATAAGCGACACNGG	TTttCtaGatTttcgGgaAaAAA	NC_071671.1 5197491	+	2	0
X	TTCACCCGGATAAGCGACACNGG	TTtctagattTtcGgGAaAaAAT	NC_071671.1 5197492	+	1	0
X	TTCACCCGGATAAGCGACACNGG	TTCtagatttTcgGgaAaAaATA	NC_071671.1 5197493	+	1	0
X	TTCACCCGGATAAGCGACACNGG	TctAgattttcggGaaAaAaTAA	NC_071671.1 5197494	+	1	0

I also noticed that these off-targets above don't have a PAM (NGG), but some others do:

#Bulge type	crRNA	DNA	Chromosome	Position	Direction	Mismatches	Bulge Size
X      TTCACCCGGATAAGCGACACNGG	cctttggGaAagttttttttGGG	NC_016056.1 15617	+	3	0
X	TTCACCCGGATAAGCGACACNGG	TTtcgggGttTtttaccaAtTGG	NC_016056.1 15484	+	4	0
X	TTCACCCGGATAAGCGACACNGG	TTttttaccAattGgtcCAaTGG	NC_016056.1 15492	+	3	0

Even if I just use Cas-Offinder I got wierd results.
running cas-offinder offinder.conf G offinder.out where offinder.conf looks like this:

/path_to/genome.fasta
NNNNNNNNNNNNNNNNNNNNNGG 0 0
TTCACCCGGATAAGCGACACCGG 5

I get this:

#gRNA target strand mismatches
TTCACCCGGATAAGCGACACCGG	TTCACCCGGATAAGCGACACCGGAAAC	+	0
TTCACCCGGATAAGCGACACCGG	TgtAaCCtcATAAGCGACACCGGTGTC	+	5
TTCACCCGGATAAGCGACACCGG	aTCACtCGGAgAAGgGACAgCGGCTGT	+	5
TTCACCCGGATAAGCGACACCGG	acttCaCccgcctctGcCACaccGGGT	-	5
TTCACCCGGATAAGCGACACCGG	acttaaaccggAtatcAgACaccGGAT	-	5

Above, the last two sequences have way more than 5 mismatches

I don't know why this is happening, can you help me?

Thank you,
Dani.

[nkuperwasser]

I've come across the same issue (in cas offinder). In my case, i am using the new silicon M chips on the mac, and there seems to be an issue with the program when using the GPU option. When I redo it using the CPU (C) option, it appears to work correctly, but much slower. In the cas-design script, I think that chaging the script to use the C option may work better. At this point, I do not know enough about the background stuff to make the GPU side work, but if you're willing to have it run a bit more slowly, it will work.