BND representation

[nnariai]

Hi @fritzsedlazeck ,

I noticed that Sniffles now reports BND as default for translocation, instead of TRA in VCF. Thank you for the modification! I like the BND representation more, as is is clearly described in the VCF v4.2 the spec: https://samtools.github.io/hts-specs/VCFv4.2.pdf

I found that for some cases (STRANDS=++ and --), BND representation in Sniffles' VCF is not following the VCF v4.2. Also, the VCF spec specifies that two BND entries should appear in VCF with MATEID for a simple translocation, instead of using one BND entry with "STRANDS=XX". It would be great if you could consider modifying the BND representation (multiple BND entries for one TRA, using MATEID, instead of STRAND=XX), so that it complies more with the VCF spec.

Just to clarify, I pasted some TRA predictions with Sniffles vs. Manta v1.5.0. Please note that I used publicly available PacBio data data for Sniffles and Illumina data for Manta (both of SKBR3), so the predicted SV positions may not match exactly (but I assume two predictions refer to the same SV). With these examples, I'm trying to convince you that the Manta BND representation (below) would comply more with the VCF spec.

• Case: +/+
Sniffles
chr4 119878405 33216 N N[chr8:43843855[ . PASS PRECISE;SVMETHOD=Snifflesv1.0.9;ZMW=29;STD_quant_start=0.462910;STD_quant_stop=0.597614;Kurtosis_quant_start=5.012614;Kurtosis_quant_stop=8.825857;SVTYPE=BND;SUPTYPE=SR;SVLEN=0;STRANDS=++;RE=29 GT:DR:DV ./.:.:29

Manta
chr4 119878405 MantaBND:68981:0:1:0:0:0:0 T T]chr8:43843854] . PASS SVTYPE=BND;MATEID=MantaBND:68981:0:1:0:0:0:1;SOMATIC;SOMATICSCORE=127;BND_DEPTH=52;MATE_BND_DEPTH=56 PR:SR 85,0:72,0 43,14:51,17
chr8 43843854 MantaBND:68981:0:1:0:0:0:1 A A]chr4:119878405] . PASS SVTYPE=BND;MATEID=MantaBND:68981:0:1:0:0:0:0;SOMATIC;SOMATICSCORE=127;BND_DEPTH=56;MATE_BND_DEPTH=52 PR:SR 85,0:72,0 43,14:51,17

• Case: +/-
Sniffles
chr1 229218656 32965 N N[chr5:138549260[ . PASS PRECISE;SVMETHOD=Snifflesv1.0.9;ZMW=26;STD_quant_start=0.620174;STD_quant_stop=0.554700;Kurtosis_quant_start=4.621364;Kurtosis_quant_stop=3.680843;SVTYPE=BND;SUPTYPE=SR;SVLEN=0;STRANDS=+-;RE=26 GT:DR:DV ./.:.:26

Manta
chr1 229218656 MantaBND:18921:0:1:0:0:0:1 G GC[chr5:138549260[ . PASS SVTYPE=BND;MATEID=MantaBND:18921:0:1:0:0:0:0;SVINSLEN=1;SVINSSEQ=C;SOMATIC;SOMATICSCORE=133;BND_DEPTH=70;MATE_BND_DEPTH=60 PR:SR 81,0:88,0 41,18:77,18
chr5 138549260 MantaBND:18921:0:1:0:0:0:0 T ]chr1:229218656]CT . PASS SVTYPE=BND;MATEID=MantaBND:18921:0:1:0:0:0:1;SVINSLEN=1;SVINSSEQ=C;SOMATIC;SOMATICSCORE=133;BND_DEPTH=60;MATE_BND_DEPTH=70 PR:SR 81,0:88,0 41,18:77,18

• Case: -/+
Sniffles
chr3 141644031 33135 N ]chr8:88792669]N . PASS PRECISE;SVMETHOD=Snifflesv1.0.9;ZMW=37;STD_quant_start=0.000000;STD_quant_stop=0.235702;Kurtosis_quant_start=7.512962;Kurtosis_quant_stop=13.846639;SVTYPE=BND;SUPTYPE=SR;SVLEN=0;STRANDS=-+;RE=37 GT:DR:DV ./.:.:37

Manta
chr3 141644031 MantaBND:53870:0:1:0:1:0:0 T ]chr8:88792669]T . PASS SVTYPE=BND;MATEID=MantaBND:53870:0:1:0:1:0:1;SOMATIC;SOMATICSCORE=71;BND_DEPTH=51;MATE_BND_DEPTH=45 PR:SR 99,0:63,0 167,23:338,23
chr8 88792669 MantaBND:53870:0:1:0:1:0:1 A A[chr3:141644031[ . PASS SVTYPE=BND;MATEID=MantaBND:53870:0:1:0:1:0:0;SOMATIC;SOMATICSCORE=71;BND_DEPTH=45;MATE_BND_DEPTH=51 PR:SR 99,0:63,0 167,23:338,23

• Case: -/-
Sniffles
chr4 87565959 33204 N ]chr19:58533693]N . PASS PRECISE;SVMETHOD=Snifflesv1.0.9;ZMW=10;STD_quant_start=1.760682;STD_quant_stop=0.836660;Kurtosis_quant_start=-0.429761 ;Kurtosis_quant_stop=0.877551;SVTYPE=BND;SUPTYPE=SR;SVLEN=0;STRANDS=--;RE=10 GT:DR:DV ./.:.:10

Manta
chr4 87565959 MantaBND:66424:0:1:0:1:0:0 T [chr19:58533694[T . PASS SVTYPE=BND;MATEID=MantaBND:66424:0:1:0:1:0:1;CIPOS=0,1;HOMLEN=1;HOMSEQ=T;SOMATIC;SOMATICSCORE=72;BND_DEPTH=51;MATE_BND_DEPTH=58 PR:SR 86,0:85,0 32,7:50,8
chr19 58533693 MantaBND:66424:0:1:0:1:0:1 A [chr4:87565960[A . PASS SVTYPE=BND;MATEID=MantaBND:66424:0:1:0:1:0:0;CIPOS=0,1;HOMLEN=1;HOMSEQ=C;SOMATIC;SOMATICSCORE=72;BND_DEPTH=58;MATE_BND_DEPTH=51 PR:SR 86,0:85,0 32,7:50,8

Thank you for taking a look!

[fritzsedlazeck]

Thanks for letting me know. Indeed for the BND I just converted the TRA record. So what you are saying is that each of the reported BND in Sniffles should be reported as two entries?

Thanks
Fritz

[nnariai]

Yes. For the cases for STRANDS=+- and -+, it would be a easy fix. However, for STRANDS=++ and --, I believe the BND representations (e.g. bracket orientation and nucleotide placement) should be fixed as per "5.4 Specifying complex rearrangements with breakends" in the vcf spec (https://samtools.github.io/hts-specs/VCFv4.2.pdf).

[fritzsedlazeck]

Thanks I will take a look as soon as possible.
Fritz

[cmdcolin]

I'd add a +1 to this

Note that the current way that is reported with the END field representing the location on the CHR2 will cause some havoc for tabix indexing because tabix uses the END= field in INFO to make it's binning index. Similar xref samtools/htslib#428

[fritzsedlazeck]

We just pushed a completely new version of Sniffles (v2.0.0.2) and hope this will be resolved.

[vmukhina]

Hi @fritzsedlazeck , this problem is still actual in new sniffles versions. I tested 2.0.2 and 2.0.7 and both give me BND representation without mateid like
chr1 8595284 Sniffles2.BND.B7B6S0 N ]chr14:87478116]N 60 PASS PRECISE;SVTYPE=BND;SUPPORT=6;COVERAGE=30,26,33,34,32;STRAND=+-;NM=0.060;AF=0.194;CHR2=chr14;STDEV_POS=1.732 GT:GQ:DR:DV 0/0:9:25:6
I also don't see any complementary entries in files and there are no MATEID flags at all.
Please let me know if there is another version with fix. Thanks!

[xsvato01]

Hi, I am reporting what I believe is the same issue. I manually looked up what should be mate pairs in vcf:

20	1102330	Sniffles2.BND.122S13	N	[2:129432299[N	60	PASS	PRECISE;SVTYPE=BND;SUPPORT=1;COVERAGE=0,0,1,1,1;STRAND=+;AF=1.000;CHR2=2;STDEV_POS=0
2	129434443	Sniffles2.BND.4F5S1	N	N[20:1102329[	60	PASS	PRECISE;SVTYPE=BND;SUPPORT=1;COVERAGE=2,1,1,1,1;STRAND=+;AF=1.000;CHR2=20;STDEV_POS=0

This BND got only one supporting read, upon clicking on the primary and supplementary mapping sites in IGV you get this:

In both, IGV and the vcf, you get all strands "+", however, according to the ALT encoding it would mean rather this:

For the variants to be mate pairs, the ALT should be in the format ]xxx]N + N[xxx[.

Please correct me if I am wrong