-
Notifications
You must be signed in to change notification settings - Fork 0
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Missing translocations when running NAIBR with candidate lists #10
Comments
Hi Christoph! Thank you for the encouraging words! It is great to hear that someone else is benefiting from this :) With (1) it seems it was indeed a bug, or more a misstake on my part as I had not considered calling translocation from candidates in my rewrite. I added a fix which should sort this out, see 778a702. Please check it out and see if it works for you too! As for (2) this is also something that has been bugging me for a while. Discordant reads are definitely present as they are used to generate a candidate SV which is then scored. However it seems this discordant reads is not counted for the final call for some reason. I have to investigate this further... |
Thank you very much for the quick reply and the immediate fix! I ran the updated version on my samples and it works! The translocations now appear in the output as expected. Great! The missing translocations were my main worry. Concerning (2): That's good to know. If the discordant reads are used for generating and scoring the candidate, and it's pretty much only a problem with the final output file, it's fine by me for now. If I need the number of discordant pairs, I can just as well use the output from generating my candidate lists. Thanks again, I really appreciate your help! |
Great to hear that you got the expected translocation! I will make a separate issue with (2) so it can be worked on in the future. |
Hi,
first of all, thank you for continuing the work on NAIBR and keeping it up to date! That's great and really appreciated!
I think NAIBR is a well though-out and promising approach, and one of the few published SV callers for 10x linked-read data, that produces reasonable results on my samples (without errors!) and is still updated. One thing I really like is that it works with candidate breakpoint lists. But I think there might be a bug when calling translocations from these candidate lists:
(1) For example, when calling the translocation in your test bam file "HCC1954T_10xGenomics_chr_12_20_TRA.bam" without candidate list, the translocation is called perfectly fine (default parameters):
Output in NAIBR_SVs.reformat.bedpe:
But when I put the previously detected breakpoints into a candidate bedpe file like this:
chr12 40393618 40393619 chr20 53974756 53974757 --
it's not detected anymore:
And that's the same with my 10x linked-read data. I have candidates from another caller, and deletions, inversions and tandem duplications are called, but never any translocation (and I know that there are some translocations in my samples). Strangely, that part worked with the old NAIBR from the Raphael-group git (but it produces other errors on some of my samples....).
And, if you don't mind, I have another question, where I'm not sure whether there's something wrong, or if I don't understand it correctly:
(2) Very often in the NAIBR output, I see SVs with a lot of split-molecules (high-score) but without any discordant read pairs. But I am sure from generating my candidates, that there are often plenty discordant reads around the breakpoint. This is not really a big problem for me, the SV call and orientation usually seems right, and I get that there are surely differences in calling the discordant read pairs. But I am asking myself: How does NAIBR infer the SVs orientation, if it didn't find any discordant reads? The orientation is often right, so maybe that's just a bug in the output?
I experienced both these issues with NAIBR 0.5 and 0.5.1, using default parameters in the config file and candidate lists from our in-house SV caller (which uses only discordant read pairs and split-reads).
I hope I could make myself clear enough. I'd really like to use your current version of NAIBR on my data and translocations are an important part in my analysis. Let me know, if you need any more info or small examples from my data for debugging.
Thanks for your work and help!
Best regards,
Christoph
The text was updated successfully, but these errors were encountered: