remove restriction of 10 alleles max in multiallelic sites

[akotlar]

• Simplifies makeHetHomozygotes and removes the rune-by-rune reading in favor of string by string, which enables us to have unlimited numbers of alleles. This is useful because dbSNP VCF format has many submissions with > 9 alleles

Previously we used a fast parsing strategy, which took a string like "0|1" and read it as runes (ascii code points) '0', '|', and '1'. This allowed us to avoid splitting on "|" and to also allow us to not split on ":" should there be metadata about the genotypes (e.g. GT:AD).

Now, we do the necessary splitting, and evaluate each allele as a string, which may be composed of multiple runes, allowing genotypes like "100|0" to be evaluated (for a hypothetical site with >= 100 alternate alleles).

We still keep a fast path for biallelic sites, entirely avoiding the need to do any processing besides exact match on the most common genotypes (0|0, 0/0, 1|0, 1/0, 0/1, 0|1).

Addresses #7

[wingolab]

I see there's a test case for the multiallelic with > 10 sites. Logic seems fine.

FYI, I needed to do the following to test it locally.

go mod init github.com/bystrogenomics/bystro-vcf
go mod tidy
go test -v ./...

[cristinaetrv]

Is the new version going to represent haploid genotypes as haploid or still homozygous?

[akotlar]

It will still be homozygous; in the for loop, each genotype value adds 1 count to the genotype count; if that genotype value matches the alleleNum (the multi allelic allele in the decomposed multi allelic, where alleleNum will be the only ALT allele on that line in the output of bystro-vcf), then alt also gets a count of 1.

• bystro-vcf/main.go

  Line 1017 in dfb9065

  gtCount++

In the haploid case, if the only genotype value matches the alleleNum, that site will be homozygous/hemizygous, since altCount == 1 and gtCount == 1

[akotlar]

I see there's a test case for the multiallelic with > 10 sites. Logic seems fine.

FYI, I needed to do the following to test it locally.

go mod init github.com/bystrogenomics/bystro-vcf
go mod tidy
go test -v ./...

Ah, I forgot to add the go.mod and go.sum, thanks

[cristinaetrv]

Lgtm!