Import of complex variants with reference overlap

It can be specified that variants are represented using reference overlap by adding the line "##refOverlap=true" in the VCF header. If no such line is found in the header, the default is "false", i.e., that no reference overlap alleles are present that need to be replaced by overlapping alleles. Note: Filtered variant positions may interfere in reference overlap processing and should be removed from VCF files prior to reference overlap import.

Detection of complex regions

When reading a reference overlap VCF file, a complex region is initiated when overlapping alleles are called on different VCF lines. Complex regions can contain hundreds of complex variants, for example if one allele has a long deletion. Alleles overlap if they share a reference nucleotide position. Insertions overlap non-insertion if they are positioned internally, not if they are positioned at either boundary.

Replacing reference overlap alleles in complex regions

For each position with a complex alternate allele, a number of placeholder reference overlap alleles (refoPloidy) are expected to be present, so that the total number of alleles in the genotype field is equal to the ploidy at that position in the sample genome. For each such position in the complex region, it is then determined how many reference overlap alleles are replaced by overlapping alternate and reference alleles (numReplaced). If any reference overlap alleles remain, they are assigned the allele depth: newAD=origAD*(refoPloidy-numReplaced)/refoPloidy, where origAD is the original allele depth for all reference overlap alleles at the position. In the "Reference overlap and depth estimate" example above (Table 2), the allele depth of the re-imported reference variant will be: newAD=6*(2-1)/2=3. In the "Reference overlap" example above (Table 2), no reference overlap alleles will remain (numReplaced=2).

Alternative import of "Reference overlap" representation

The method above can be used for both "Reference overlap" and "Reference overlap with depth estimate" representations. However, a VCF file generated with the "Reference overlap" representation can also be imported correctly by simply importing as if it has no reference overlap, and subsequently removing all reference alleles with zero CLCAD2 allele depth.