Estimate a Conditional Second-Genome Share from Low-Pass Counts
Source:R/count_mixture_lowpass.R
rduckhts_count_mixture_lowpass.RdUses first-read and discordant-pair moments with fixed external calibration.
This estimator is separate from rduckhts_count_error_fit(). It is suitable
for low-pass, cell-free DNA or low-input DNA counts only under the supplied
population frequencies and calibration; the share is not an unconditional
biological fraction. Matched-frequency unrelated and IBD1 hypotheses can
have identical moments and are reported as unresolved.
Usage
rduckhts_count_mixture_lowpass(
con,
counts_table,
seq_error,
artefact_weight,
homozygosity_excess,
block_bases = 1e+07,
max_depth = 8,
min_blocks = 30,
max_blocks = 512,
max_candidates = 16,
max_state_bytes = 2^28,
table_name = NULL,
overwrite = FALSE
)Arguments
- con
A DuckDB connection with DuckHTS loaded.
- counts_table
Table or view containing
sample_id,candidate,chrom, one-basedpos,ref_count,alt_count,other_count,sample_afandcandidate_af. Counts andsample_afmust agree across candidates; site keys must be unique within each sample and candidate. Frequencies refer to the counted alternate allele. Sites missing any candidate frequency are unavailable to every candidate of that sample.- seq_error
Required fixed sequencing error in
[0, 0.5).- artefact_weight
Required fixed uniform-count artefact weight in
[0, 1).- homozygosity_excess
Required fixed excess of homozygosity in
[0, 1].- block_bases
Bases per chromosome-local block, at least 1.
- max_depth
Maximum ref plus alt depth, 1 to 65535. Deeper sites are excluded;
other_countcontributes only to QC totals.- min_blocks
Informative blocks required for 95 percent limits, at least 2, default 30. Matched frequencies require pair-bearing blocks; otherwise read-bearing blocks count. Lower values support small-fixture checks, not well-calibrated inference. Below this floor the share is reported but limits are NA and status is
insufficient_blocks.- max_blocks
Maximum physical blocks per sample, 1 to 4096.
- max_candidates
Maximum candidates per sample, 1 to 64.
- max_state_bytes
Maximum bytes of concurrently owned native moment buffers and fit copies. DuckDB owns SQL sorting, materialization and spill.
- table_name
Optional output table; otherwise return a data frame.
- overwrite
Whether to replace an existing output table.
Value
One row per sample, candidate and relationship model. IBD1 rows
require exact candidate_af == sample_af at every input site, including
zero-read and excluded sites. Columns include the bounded share (0 to 0.5),
block-jackknife standard error and score-inverted limits, singles share,
pair residual, relationship diagnostic, QC read/pair totals, calibration,
score rank and status. The discordance ratio divides discordant pairs by
SUM(choose(depth, 2) * sample_af * (1 - sample_af)).
When table_name is supplied, returns invisible TRUE.