SV not called

[ielvers]

Hi,

In case 26025, I noticed a heterozygous deletion on the chromograph. It's also clearly visible in GENS. But the deletion does not appear to be called. I don't find it when looking for all SVs on chr 13, or when searching for specific (disease associated) genes that are deleted.

Chromograph:

GENS:

This variant is called as homozygous, but it's actually hemizygous because the entire gene is deleted on one allele:
https://scout.scilifelab.se/cust003/26025/dc6e33faaf95f7bdd040a86894e79576

I don't understand how this large deletion managed to go undetected.

Thanks,
Ingegerd

[dnil]

Thank you for reporting! The most typical would be for it to be too common to score well enough for loading, while being rare enough not to end up affecting the Gens PoN too much. But let's see!

[dnil]

Mm, the breaks are in LTRs, but it is kind of called by Manta, only it gets the direction wrong somehow:

13      23518541        MantaDUP:TANDEM:93786:1:2:0:0:0 C       <DUP:TANDEM>
...
;SVLEN=1421932;SVTYPE=DUP;clingen_ngiAF=0.21;clingen_ngiOCC=61;clinical_genomics_loqusFrq=0.05649;clinical_genomics_loqusObs=1023;clinical_genomics_m
ipAF=0.2463;clinical_genomics_mipOCC=546;manta_CHROM=MantaDUP_TANDEM_93786_1_2_0_0_0|13;manta_FILTERS=MantaDUP_TANDEM_93786_1_2_0_0_0|PASS;manta_INFO=MantaDUP_TANDEM_93786_1_2_0_0_0|END:24940473|SVTYPE:DUP|SVLEN:1421932|IMPRECISE

That one is pretty common as well (20% in SweGen, 5% in the local db).

[dnil]

All in all rank score -1, and not loaded. So question is why manta calls it as a DUP. It is also odd that we don't see a CNVnator-call there. I can see TIDDIT thinking it a false positive with the noise on the breakpoints, but still not great for it to miss it either. Lets pass it on to the pipeline developers to chew on!

[jemten]

Hmm, we do have this variant in the non filtered file that looks to be matching quite well to me. Trying to find out where it went

13      23507757        SV_100_1:tiddit|CNVnator_del_6:cnvnator|MantaDEL:79515:0:1:0:2:0:manta  N       <DEL>   50      PASS    SVTYPE=DEL;SVLEN=1422066;END=24929823;REGIONA=23507408,23507894;REGIONB=24929823,24930197;LFA=9,2;LFB=8,4;LTE=6,1;CTG=.;uniquedragon_tiddit_ACC19847A11_CHROM=SV_100_1|13;uniquedragon_tiddit_AC
C19847A11_POS=SV_100_1|23507757;uniquedragon_tiddit_ACC19847A11_QUAL=SV_100_1|50;uniquedragon_tiddit_ACC19847A11_FILTERS=SV_100_1|PASS;uniquedragon_tiddit_ACC19847A11_SAMPLE=SV_100_1|ACC19847A11|GT:0/1|CN:1|COV:38:15.513722662131368:41|DV:6|RV:1|LQ:0.030303030303030304:0.25748502994011974|RR:26:2|DR:18:8;uniquedragon_t
iddit_ACC19847A11_INFO=SV_100_1|SVTYPE:DEL|SVLEN:1422066|END:24929823|REGIONA:23507408:23507894|REGIONB:24929823:24930197|LFA:9:2|LFB:8:4|LTE:6:1|CTG:.;SUPP_VEC=1;IMPRECISE;natorRD=0.503199;natorP1=1.12076e-13;natorP2=0;natorP3=1.12234e-13;natorP4=0;natorQ0=0.0029489;SVINSLEN=1;SVINSSEQ=A;VARID=CNVnator_del_6:cnvnator|
MantaDEL:79515:0:1:0:2:0:manta;set=Intersection;FOUNDBY=3;tiddit_CHROM=SV_100_1|13;cnvnator_CHROM=CNVnator_del_6|13;manta_CHROM=MantaDEL_79515_0_1_0_2_0|13;tiddit_POS=SV_100_1|23507757;cnvnator_POS=CNVnator_del_6|23508001;manta_POS=MantaDEL_79515_0_1_0_2_0|23507737;tiddit_QUAL=SV_100_1|50;cnvnator_QUAL=CNVnator_del_6|.
;manta_QUAL=MantaDEL_79515_0_1_0_2_0|654;tiddit_FILTERS=SV_100_1|PASS;cnvnator_FILTERS=CNVnator_del_6|PASS;manta_FILTERS=MantaDEL_79515_0_1_0_2_0|PASS;tiddit_SAMPLE=SV_100_1;cnvnator_SAMPLE=CNVnator_del_6;manta_SAMPLE=MantaDEL_79515_0_1_0_2_0;tiddit_INFO=SV_100_1|SVTYPE:DEL|SVLEN:1422066|END:24929823|REGIONA:23507408:2
3507894|REGIONB:24929823:24930197|LFA:9:2|LFB:8:4|LTE:6:1|CTG:.|SUPP_VEC:1;cnvnator_INFO=CNVnator_del_6|END:24930000|SVTYPE:DEL|SVLEN:-1422000|IMPRECISE|natorRD:0.503199|natorP1:1.12076e-13|natorP2:0|natorP3:1.12234e-13|natorP4:0|natorQ0:0.0029489;manta_INFO=MantaDEL_79515_0_1_0_2_0|END:24929971|SVTYPE:DEL|SVLEN:-14222
34|SVINSLEN:1|SVINSSEQ:A;svdb_origin=tiddit|cnvnator|manta;SUPP_VEC=111 GT:CN:COV:DV:RV:LQ:RR:DR        0/1:1:38,15.5137,41:6:1:0.030303,0.257485:26,2:18,8
```

[dnil]

Some svdb/other collision with the overlapping "MantaDUP"?

[jemten]

Perhaps. We have removed all the intermediate files in order to save disk space on hasta so troubleshooting is a bit of a pain

[jemten]

Sorry for the long wait.
So the reason that you don't see the call is that it is filtered out due to being too common. SVDB matches the variant to this variant from swefreq:

13  23507479    cluster_187698  N   <DEL>   .   PASS    SVTYPE=DEL;END=24929601;SVLEN=1422122;NSAMPLES=949;OCC=725;FRQ=0.763962065332

and since it has a frequency of over 70% in swefreq it gets filtered out.

[dnil]

The next question then becomes where that db cluster came from? 🤔 I don't think we have the SV calls very accessible on the swefreq site anymore, but the average coverage there seems way fine for having 75% deletions? Not to speak of the number of recessive SGCG limb-girdle dystrophies we would have in the population. 😊

[ielvers]

Thanks for digging into this Anders. I agree with Daniel though. This deletion doesn't show up at all in the gnomAD v2 SV dataset and there are 12 heterozygotes in gnomAD v4. So it can't be that common?

[jemten]

It is a bit odd. The swegen data has few very high frequency variants in this region.

13      23506905        cluster_53784   N       <DEL>   .       PASS    SVTYPE=DEL;END=24928953;SVLEN=1422048;NSAMPLES=997;OCC=381;FRQ=0.382146439318
13      23506932        cluster_187699  N       <DEL>   .       PASS    SVTYPE=DEL;END=24928963;SVLEN=1422031;NSAMPLES=949;OCC=722;FRQ=0.760800842993
13      23507211        cluster_202547  N       <DUP>   .       PASS    SVTYPE=DUP;END=24929399;SVLEN=1422188;NSAMPLES=949;OCC=847;FRQ=0.892518440464
13      23507328        cluster_56620   N       <DUP>   .       PASS    SVTYPE=DUP;END=24929482;SVLEN=1422154;NSAMPLES=997;OCC=380;FRQ=0.381143430291
13      23507446        cluster_53783   N       <DEL>   .       PASS    SVTYPE=DEL;END=24929600;SVLEN=1422154;NSAMPLES=997;OCC=382;FRQ=0.383149448345
13      23507479        cluster_187698  N       <DEL>   .       PASS    SVTYPE=DEL;END=24929601;SVLEN=1422122;NSAMPLES=949;OCC=725;FRQ=0.763962065332

[dnil]

Ping @J35P312 💯

[47KW]

Finns i DGV så filtreras bort som vanlig, case closed?

[J35P312]

Vi borde kolla hur swegen cnvnator ser ut, är den avropad av cnvnator (dvs på läsdjup) eller enbart av diskordanta par (förmodligen en alu eller liknande som klassats fel av manta). Kanske ett tecken på att vi borde ha separata databaser per caller?

[47KW]

bäst jag visar "loss", de röda