How does CRISPResso2 handle deletions between two gRNAs

[GreenSeaBug]

Hello,

I am wondering how does CRISPResso2 handle/report deletion edits between two gRNAs when there are also small indel edits at each target site? Can anyone share a screen shot of this type of result?

Thank you

[Colelyman]

Hi @GreenSeaBug,

Good question, I will try and work on an example to show you but for now here is a verbal explanation. Assuming that you are using the default parameters, there would be two quantification windows that are +/- 1 centered at -3 of both guides. Because there are edits at at least one target site, this read would be classified as edited. If you only had deletions between the guides, then that read would be unedited. Does that make sense?

Thanks,
Cole

[Colelyman]

Here is a small example dataset showing the behavior. There are two guides, and there are several examples of small indels at the target site and in between the two guides. Note that there is one read that has an insertion between the two guides (and no edits at the target sites) and it is classified as unmodified.

Attached is the input file with the edits described as the read names. And here is the command to reproduce/play around with:

CRISPResso -r1 inputs/FANC.Cas9.fastq -a CGGATGTTCCAATCAGTACGCAGAGAGTCGCCGTCTCCAAGGTGAAAGCGGAAGTAGGGCCTTCGCGCACCTCATGGAATCCCTTCTGCAGCACCTGGATCGCTTTTCCGAGCTTCTGGCGGTCTCAAGCACTACCTACGTCAGCACCTGGGACCCCGCCACCGTGCGCCGGGCCTTGCAGTGGGCGCGCTACCTGCGCCACATCCATCGGCGCTTTGGTCGG -g GGAATCCCTTCTGCAGCACC,CCGAGCTTCTGGCGGTCTCA --fastq_output --plot_window_size 40 --place_report_in_output_folder --debug

Hope this is helpful!
Cole

9.Alleles_frequency_table_around_sgRNA_CCGAGCTTCTGGCGGTCTCA.pdf
9.Alleles_frequency_table_around_sgRNA_GGAATCCCTTCTGCAGCACC.pdf
FANC.Cas9.fastq.gz

[Colelyman]

[GreenSeaBug]

Thanks for your reply, Colelyman. Apologies, I don't think I was clear in my original post. I understand that if there is no edit in the quantification window then the read will be classified as not edited. However, I was trying to describe an edit where there is complete drop out of the sequence between the two guides (i.e. a big deletion). This could be a perfect drop out (no indel at the drop out junction) or imperfect dropout (indel at the junction), but the dropout would overlap the quantification window. Since the dropout would overlap the quantification window, I presume the read would be classified as edited. But would it explicitly classify the read as containing a drop out? Or would it just say there is a read with a large deletion and one would have to use a large plot_window_size to see that it is indeed a drop out?

[GreenSeaBug]

Ah, yes of course, you are quite right! I used amplicon_min_alignment_score 0 and that solved the problem. Thank you!

However I was not able to see the large deletion in the plot window as I always got an error message when I tried to increase the plot window size... "ERROR: Offset around cut would be greater than reference sequence length. Please decrease plot_window_size parameter. Cut point: 320 window: 80 reference: 397

[Colelyman]

Glad to hear that setting a lower minimum alignment score fixed your problem!

Would you mind sharing how long your amplicon is and if you used a plot_window_size parameter?

[GreenSeaBug]

I have three reference alleles: 397 bp, 393 bp, 392 bp. I did use the plot_window_size parameter and found that the error message popped up when I increased it to 80 or above. The deletion I'm looking for is 196 bp.

[Colelyman]

Currently the plot window size can't extend past the end of the amplicon. In your example, the cut point is centered at 320, and given a window of 80, it would extend past the end (320 + 80 = 400, 400 > 397).

Probably the best way to find your deletion would be in the Alleles_frequency_table.txt file (it is in the Alleles_frequency_table.zip file). In this file there is even a column with the number of deletions so you can filter for your 196 bp deletion. Additionally, you can try the --write_detailed_allele_table parameter which will add more information into that file (like the positions of deletions, insertions, and substitutions).

[GreenSeaBug]

OK thank you!

[GreenSeaBug]

In my tests I am finding that reads with a dropout deletion between the two target sites will not map to the reference, and nor are they classified as ambiguous.