Somatic-filtered files are intermediate?

[rhughwhite]

I noticed that the filtered outputs are considered as intermediate e.g. /hot/project/disease/HeadNeckTumor/HNSC-000084-LNMEvolution/pipelines/call-sSV/TOP_UP/call-sSV-5.0.0/ILHNLNEV000003-T001-P01-F_realigned_recalibrated_merged_dedup/DELLY-1.1.3/intermediate/DELLY-1.1.3_HNSC0000016_ILHNLNEV000003-T001-P01-F-realigned-recalibrated-merged-dedup_somatic-filtered.bcf

In many cases will these files not be the final outputs users want?

[Faizal-Eeman]

The final BCF in the ./output/ has SVs that passed the variant filtering criteria FILTER=='PASS'. So if a user wants to look at all variants (PASS + non-PASS), can use the *somatic-filtered.bcf from ./intermediate/.

[rhughwhite]

Ah my mistake. The reason I ask is because Lydia noticed differences in the number of variants reported by v5.0.0 relative to v1.0.0, and we thought it may be due to filtering since in v5.0.0 the file names don't contain the phrase filtered.

But it seems that this is probably due to changes between Delly 0.8.7 and 1.1.3. The newer version reports more variants, though this is not necessarily a problem, what do you think @Faizal-Eeman?

From @lydiayliu

Delly 1.1.3 called SVs (had to first convert bcf to vcf):
yiyangliu@ip-0A125216:/hot/project/disease/HeadNeckTumor/HNSC-000084-LNMEvolution/data/Delly/sSV/topup/DELLY-1.1.3$ for a in *bcf.vcf; do echo ${a}; cat ${a} | grep -v '^#' | wc -l; done
DELLY-1.1.3_HNSC0000016_ILHNLNEV000003-T001-P01-F-realigned-recalibrated-merged-dedup.bcf.vcf
125
DELLY-1.1.3_HNSC0000016_ILHNLNEV000003-T002-L01-F-realigned-recalibrated-merged-dedup.bcf.vcf
110
DELLY-1.1.3_HNSC0000016_ILHNLNEV000003-T003-L02-F-realigned-recalibrated-merged-dedup.bcf.vcf
53
DELLY-1.1.3_HNSC0000016_ILHNLNEV000014-T001-P01-F-realigned-recalibrated-merged-dedup.bcf.vcf
59
DELLY-1.1.3_HNSC0000016_ILHNLNEV000014-T002-L01-F-realigned-recalibrated-merged-dedup.bcf.vcf
72
DELLY-1.1.3_HNSC0000016_ILHNLNEV000014-T003-L02-F-realigned-recalibrated-merged-dedup.bcf.vcf
51

Delly 0.8.7 FILTERED SVs:
yiyangliu@ip-0A125216:/hot/project/disease/HeadNeckTumor/HNSC-000084-LNMEvolution/data/Delly/sSV/topup$ for a in *bcf.vcf; do echo ${a}; cat ${a} | grep -v '^#' | wc -l; done
filtered_somatic_ILHNLNEV000003-T001-P01-F_realigned_recalibrated_merged_dedup.bcf.vcf
120
filtered_somatic_ILHNLNEV000003-T002-L01-F_realigned_recalibrated_merged_dedup.bcf.vcf
102
filtered_somatic_ILHNLNEV000003-T003-L02-F_realigned_recalibrated_merged_dedup.bcf.vcf
24
filtered_somatic_ILHNLNEV000014-T001-P01-F_realigned_recalibrated_merged_dedup.bcf.vcf
12
filtered_somatic_ILHNLNEV000014-T002-L01-F_realigned_recalibrated_merged_dedup.bcf.vcf
9
filtered_somatic_ILHNLNEV000014-T003-L02-F_realigned_recalibrated_merged_dedup.bcf.vcf
3

[Faizal-Eeman]

But it seems that this is probably due to changes between Delly 0.8.7 and 1.1.3. The newer version reports more variants, though this is not necessarily a problem, what do you think @Faizal-Eeman?

Yes, that is correct. Although, all variants called by DELLY 0.8.7 should also have been called by 1.1.3.

And the extra variants called by 1.1.3 could be calls that have

• low MAPQ calls
• low quality score

or because of

• overlapping SVs: same start position but alternating end position & vice versa

See Discussion #39 on Low MAPQ variants.

[rhughwhite]

Thanks for the clarification! Closing this.