'Too few reads' even if '--min_support 1' about sniffles HOT 17 CLOSED

Hi
Sniffles was designed to scan reads. In the first step it randomly picks reads to estimate the error rates and other parameters. That is where you get the error from as we cannot estimate these parameters over 3 contigs.

I can implement an option for that later next week. (Sorry I am about to travel to a conference).
Cheers
Fritz

from sniffles.

@fritzsedlazeck

The raw pacbio fastq and corrected pacbio fastq have different error rates. Should we add a option for this if we skip the error rate estimating?

maybe the corrected pacbio fastq is better for big SV.

from sniffles.

Thanks for pinging me again.
I can add the skip parameter. It is just that I would need an estimation for the parameters estimated....

About error corrected reads. I still have to benchmark if with the error corrected reads you are risking to loose some e.g. het indels. I am currently using uncorrected reads.
However, I have heard that you get a more robust signal (e.g. breakpoints) with the error corrected reads.

Sorry for the delay
Fritz

from sniffles.

Hi
I just added the option and pushed everything. Please give it a try and let me know how it goes.
thanks
Fritz

from sniffles.

The command works,
but no SV are outputted to vcf file. a DEL and a BND are expected.

sniffles --report_BND --skip_parameter_estimation --min_support 1 -m bnd.sort.bam -v bnd.vcf

test.sv.zip

from sniffles.

Sorry to hear that. I have just tried with a normal read set and there I got the calls.
Could you share your example? ([email protected])
If its too large I can open up a dropbox.

As you know its not really the focus for Sniffles, but it should work...
Thanks
Fritz

from sniffles.

@fritzsedlazeck

I have shared the test.sv.zip in above(3 days ago).

Yes, It is not really the focus for Sniffles, but it should work. And I want to test it with a small test case.

from sniffles.

got it sorry I overlooked that.

from sniffles.

Oh I see why its not working. you used minimap2 , which does not report the NM tag. Thus Sniffles ignores these reads.

Did you try NGMLR? That is currently recommended.
I will at some stage include the support for minimap2 but that is currently not straight forward.

from sniffles.

@fritzsedlazeck

Sniffles +NGMLR work well for DEL now.
but Sniffles +NGMLR not work for BND yet.

a sample of test.sv.BND.zip
test.sv.BND.zip

from sniffles.

@fritzsedlazeck

Sniffles +NGMLR work well for DEL now when 3 reads in the sam.
Sniffles +NGMLR not work for DEL when 1 reads in the sam.

a sample of test.svn.del.zip.

test.sv.DEL.zip

from sniffles.

Thanks I will take a look. Probably something related to post filtering.

from sniffles.

Hi,
I checked with the del example that you sent that seems to work fine for me. In the DEL zip your vcf file also shows the del for 1 read.

I had to correct one bug for the TRA example to make it work with only 1 read support. (there was a >1 in the code for TRA)
Please clone and recompile the code and let me know if that solves it.

Thanks
Fritz

from sniffles.

It works well now. thanks a lot for your job.

from sniffles.

@fritzsedlazeck

I checked the DEL report again.

1. some small DEL near the head/tail of the read maybe reported as soft-clip by aligner.
   in this case, no del will be report by sniffles too.

2. DEL such as '4S994M999D135675M1S' will be reported well.
   but some DEL such as 4S496M99D137074M will not be reported by sniffles.
   in this case, '--del_ratio 0.00001 --ins_ratio 0.00001' not work too.
   which rule it is filtered by?
   test.sv.del.rev2.zip is an example of '4S496M99D137074M'
   test.sv.del.rev2.zip

from sniffles.

I have updated the code. Please clone and try it out.
Thanks for providing all these test cases!
Cheers
Fritz

from sniffles.

It works well now.

from sniffles.