Skip to content

ug_hcr.md

Latest commit

 

History

History
56 lines (42 loc) · 3.72 KB

ug_hcr.md

File metadata and controls

56 lines (42 loc) · 3.72 KB

UG-High confidence region

Version Log

  • v1.0 Initial version.
  • v1.1 Fixed a mistake in annotation of the exclusion regions.
  • v1.2 ug_cnv_lcr added. see description below.
  • v1.2.1 annotation tag updated per
  • v1.3 HMER11 was split off GAP annotation (that annotates Ns in the reference +- 4 bases)
  • v1.4 ug_cnv_lcr updated: annotation tags were merged and additional regions (telomere-centromere) were added.
  • v2.1 UG-HCR updated for the data from RM6 ePCR formulation and basecalling 5.0.
  • v2.1.1 ug_cnv_lcr updated: telomere annotation tags were updated to include chromosome edges.
  • v2.1.2 sorted bed files based on chromosome order in reference genome
  • v3.0 ug_hcr is developed based on gVCFs of multiple runs.
  • v3.1 ug_hcr was updated so that the chrY HCR is defined based on male samples only.

Definition of ug_lcr.bed and ug_hcr.bed

  1. Genomic areas where UG performance is consistently of lower quality are annotated in ug_lcr.bed. The regions were defined as follows:
  2. gVCFs were generated from 56 GIAB samples sequenced to a coverage of ~40X (HG001-7).
  3. Run HCR was defined as any block (reference or variant) that has GQ>=20
  4. Run HCRs were intersected and any block that occured in less than 35 samples was assigned to ug_lcr.bed
  5. On chrY the HCR was defined requiring intersection of at least 21 male samples' HCRs.
  6. In addition, reference homopolymers of length 15 or longer were extended by three bases to the left and two bases to the right and also assigned to ug_lcr.bed.

Total excluded BED size is 256 Mb. Majority of the excluded bases lie outside of GIAB high confidence regions where less than 1% is excluded from each GIAB sample high confidence region (version 4.2.1).

  1. ug_hcr.bed are the regions complementary to ug_lcr.bed.

Intersection of UG-HCR with various genomic regions

tag HG001 HG002 HG003 HG004 HG005 HG006 HG007 exome WG
excluded bases 21,913,516 23,639,890 23,175,722 22,500,237 19,207,155 21,368,823 21,324,483 1,014,035 256,672,961
% excluded bases 0.77 0.83 0.81 0.79 0.68 0.75 0.75 3.06 7.90

Description ug_cnv_lcr

We defined regions where coverage biases and mappability issues are likely to generate noisy copy number calls (UG-CNV-LCR). These regions were defined by the following criteria based on the coverage profile of 90 unrelated samples sequenced to approximate coverage of 50x on the UG sequencer.  

  • Coverage-Mappability: 
    • Low mappability - defined as 50 bp windows that, on average, are covered by at least 20 reads, but less than 10% of these reads are aligned with mapping quality > 20. 
    • High coverage variability - 50 bp windows with coverage that is highly variable between samples (std/mean > 0.5) 
    • Low coverage - defined as 50 bp windows that, on average, are covered by 10 reads or less. 
  • Telomere_Centromere:  250kb at the end of each chromosome, 250kb flanking sequence around the centromeres 
  • Clusters: regions that were empirically found to generate noisy CNV calls and thus were marked in the callset.

SV calls that had larger than 50% overlap with excluded areas were marked as “UG-CNV-LCR” in the callset.  

The following table details each UG-CNV-LCR criteria by size

tag GIAB exome %GIAB %exome %hg38
Coverage-Mappability 157,351,620 42,620,663 977,296 1.7% 2.9%
Telomere-Centromere 66,802,541 5,151,741 148,638 0.2% 0.4%
Clusters 7,087,267 1,595,921 73,661 0.06% 0.2%