Skip to content

Ploidy & Special Chromosomes

AFQuery computes ploidy-aware AN for sex chromosomes (chrX, chrY) and the mitochondrial chromosome (chrM). This ensures that allele frequencies are correct when querying these chromosomes, where the number of alleles per sample differs from the diploid autosomes.

Chromosome name normalization

AFQuery accepts MT, chrMT, and chrM as input; output always uses chrM.


Ploidy Rules

Chromosome Female AN contribution Male AN contribution
Autosomes (chr1–22) 2 2
chrX (non-PAR) 2 1
chrX (PAR1, PAR2) 2 2
chrY (non-PAR) 0 1
chrY (PAR1, PAR2) 0 0 — counted on chrX
chrM 1 1

For each eligible sample at a given position, AFQuery adds the appropriate ploidy count to AN based on the sample's sex and the chromosome/position.

A sample contributing ploidy 0 is not eligible at that position at all: a female has no chrY to genotype, so she is neither a carrier nor homozygous reference there. On chrY, n_eligible, N_HOM_REF and AN therefore count males only. In the chrY PARs nobody is eligible (see below).

Karyotypes

The manifest records each sample as male or female, read as XY or XX. Other sex-chromosome karyotypes (45,X; 47,XXY; mosaics) are not modelled: a sample is given the ploidy of the sex it is registered with.


Pseudoautosomal Regions (PAR)

PAR1 and PAR2 are the same sequence at the two ends of chrX and chrY, and they recombine like an autosome: a male carries one copy on his X and one on his Y, so at a PAR locus every sample is diploid. The reference genome holds that sequence twice, once on each chromosome, so AFQuery counts every PAR locus once, at its chrX coordinates:

  • chrX PAR: diploid for both sexes — males and females contribute AN=2.
  • chrY PAR: not genotyped. Nobody is eligible there, so a query returns no results and AN is 0. This keeps a PAR variant from being counted twice, and keeps a male carrier from being taken for a hemizygote.

Most analysis references hard-mask the chrY PARs (the GRCh38 analysis sets do), so variant callers report PAR variants on chrX anyway. Calls at chrY PAR coordinates only appear when the reads were aligned to an unmasked reference; they are stored, but not counted. A point query (query --locus, variant-info) at such a position warns that it lies in a PAR; region and batch queries, dump and annotate skip it silently (annotate writes AFQUERY_AN=0).

PAR coordinates by genome build:

GRCh38

chrX:

Region Start End
PAR1 10,001 2,781,479
PAR2 155,701,383 156,030,895

chrY:

Region Start End
PAR1 10,001 2,781,479
PAR2 56,887,903 57,217,415

GRCh37 / hg19

chrX:

Region Start End
PAR1 60,001 2,699,520
PAR2 154,931,044 155,260,560

chrY:

Region Start End
PAR1 10,001 2,649,520
PAR2 59,034,050 59,363,566

Positions within PAR1 or PAR2 on chrX are treated as diploid for all samples; positions within PAR1 or PAR2 on chrY are not genotyped.


Effect on AF Queries

chrY

Querying chrY with --sex female returns AN=0 (females have no Y chromosome):

afquery query --db ./db/ --locus chrY:2787758 --sex female
# chrY:2787758 — no results (AN=0 for all variants)

afquery query --db ./db/ --locus chrY:2787758 --sex male
# chrY:2787758 C>T  AC=3  AN=856  AF=0.0035  n_eligible=856  N_HET=0  N_HOM_ALT=0  N_HEMI=3  N_HOM_REF=853  N_FAIL=0  N_NO_COVERAGE=0

Inside a chrY PAR (here GRCh38 PAR1), nothing is reported, whatever the filters:

afquery query --db ./db/ --locus chrY:1500000
# AfqueryWarning: chrY:1500000 lies in a pseudoautosomal region (PAR); PAR variants
# are counted at their chrX coordinates, so nobody is genotyped here.
# No variants found for the given filters.

Query the same locus at its chrX coordinates instead (chrX:1500000).

chrX non-PAR

Male samples contribute AN=1, female samples contribute AN=2. This means a cohort of 500 females and 500 males has AN = 500×2 + 500×1 = 1500 at a non-PAR X position.

afquery query --db ./db/ --locus chrX:100000000
# N_total = 1000 samples, AN = 1500 (not 2000)

chrM

All samples are haploid at mitochondrial loci:

afquery query --db ./db/ --locus chrM:3243
# AN = n_samples (one allele per sample)

Genotype Counting

Counting Identity

For every query result at a biallelic site, the following identity holds:

N_HET + N_HOM_ALT + N_HEMI + N_HOM_REF + N_FAIL + N_NO_COVERAGE = n_eligible

This can be used to validate results. At a multi-allelic site, eligible samples that carry only another ALT allele at the position fall in none of these categories, so the sum is lower by exactly that number of samples (see Multi-allelic sites). N_HOM_REF is the number of eligible samples that carry only the reference allele and have no failed call at the position: homozygous reference where they are diploid, and, where they are haploid, hemizygous for the reference (a male on chrX non-PAR or chrY) or a non-carrier on chrM. N_NO_COVERAGE is 0 unless a coverage-evidence filter is active — see Coverage Evidence.

Mutual exclusivity

N_HET, N_HOM_ALT, N_HEMI, N_HOM_REF, N_FAIL, and N_NO_COVERAGE are mutually exclusive. A sample with a failed call for any allele at the position is counted in N_FAIL only — it does not appear in N_HET, N_HOM_ALT, N_HEMI or N_HOM_REF, and is left out of AC and AN.

The genotype counts are counted per sample, not per allele, and each follows the sample's ploidy at the position:

Count Who Copies of the ALT allele (AC)
N_HET diploid sample, GT 0/1 1
N_HOM_ALT diploid sample, GT 1/1; every carrier on chrM 2; 1 on chrM
N_HEMI male carrier on chrX non-PAR or chrY (GT 1, or 1/1 written diploid) 1

So AC = N_HET + 2 × N_HOM_ALT + N_HEMI everywhere except chrM, where AC = N_HOM_ALT. N_HEMI is always 0 on the autosomes, the chrX PARs and chrM, and for females.

chrX non-PAR

  • A male with GT=1 (or 1/1) is hemizygous: he has a single copy of the locus. He contributes AC=1, AN=1 and is counted in N_HEMI.
  • A male with GT=0/1 is a failed call: REF and ALT cannot both be the genotype of a single copy (see Genotypes inconsistent with ploidy).
  • A female with GT=0/1 contributes AC=1, AN=2 and is counted in N_HET; with GT=1/1, AC=2, AN=2 and N_HOM_ALT.

N_HET and N_HOM_ALT describe diploid genotypes only, so at a non-PAR chrX position the males are never in them: their carriers are in N_HEMI and the rest in N_HOM_REF (or N_FAIL).

chrY

chrY outside the PARs is haploid in males and absent in females (who are not eligible there):

  • Every carrier is hemizygous, counted in N_HEMI
  • N_HET and N_HOM_ALT are always 0 on chrY

chrM

chrM is haploid for all samples (both sexes contribute AN=1), but it is not hemizygous: a cell holds hundreds to thousands of mtDNA copies, and a variant may be present in all of them (homoplasmy) or in a fraction (heteroplasmy). AFQuery stores the discretized genotypes of the source VCF and records no heteroplasmy level (see the FAQ), so:

  • Every carrier counts once, in N_HOM_ALT, whatever its GT (1, 0/1 or 1/1); a 0/1 call is not an error there
  • N_HET and N_HEMI are always 0 on chrM

Genotypes inconsistent with ploidy

Ploidy comes from each sample's sex and the position, not from how the VCF writes the genotype. A male carrier on chrX non-PAR or chrY counts as one allele whether its GT is written haploid (1) or diploid (1/1): both name a single allele. A call that names more alleles than the sample has cannot be counted that way:

  • on chrX non-PAR or chrY in a male, a call naming two different alleles: a heterozygous call (0/1) or two ALT alleles (1/2). A male has a single copy there, so REF and ALT cannot both be his genotype; such a call usually comes from mis-mapped reads or contamination, and a 47,XXY sample registered as male shows it genuinely;
  • on chrM, two different ALT alleles (1/2). A 0/1 call on chrM is not inconsistent: it is heteroplasmy, a mix of mtDNA copies, and the sample is a carrier (see chrM);
  • on a diploid position, more than two copies across the records at that position (e.g. 1/1 for one ALT and 0/1 for another).

create-db and update-db --add-samples record every call of the sample at such a position as a failed call, so it counts in N_FAIL and is left out of AC and AN (counted as hom-ref with --fail-as-ref), and log a warning naming the sample and the positions.

A cohort called in diploid mode on the sex chromosomes therefore shows male N_FAIL at chrX positions where the caller saw both alleles in a male. That is the intended reading, the one gnomAD applies too: those genotypes are unknown, not hom-ref and not carriers. Do not rewrite them with bcftools +fixploidy beforehand; it keeps the first allele of each call (0/1 becomes 0), which turns them into hom-ref without a trace (see Preprocessing).


Sex Filter Interaction

When --sex female is used on chrX (non-PAR), AN is purely diploid:

  • Each eligible female contributes AN=2
  • AF is computed over a fully diploid denominator

When --sex male is used on chrX (non-PAR), AN is purely haploid:

  • Each eligible male contributes AN=1
  • AF reflects the observed allele frequency in haploid male calls
  • Carriers are all hemizygous: N_HEMI = AC, and N_HET = N_HOM_ALT = 0

This makes it straightforward to compare X-linked variant frequencies between sexes without manual ploidy adjustment.


Next Steps