Parameters for generate_haplotypes().
More...
#include <recombinator.hpp>
◆ preset_t
Preset parameters for common use cases.
| Enumerator |
|---|
| preset_default | Default parameters.
|
| preset_haploid | Best practices for haploid sampling.
|
| preset_diploid | Best practices for diploid sampling.
|
◆ scoring_model_t
Kmer scoring model used for a chain.
| Enumerator |
|---|
| standard_scoring | Standard model: absent, heterozygous, present, and frequent bands are scored according to the diploid/haploid coverage expectation.
|
| high_coverage_scoring | High-coverage model. Kmers in the frequent category (count above the homozygous threshold) contribute the present score, while all other kmers contribute the absent score. This is intended for contigs such as chrM, where the true signal is in the frequent component and the haploid/diploid peaks are contamination (e.g. NuMTs or recurrent errors).
|
| half_coverage_scoring | Half-coverage model for heterogametic allosomes. The single true copy sits at ~cov/2, which the standard model labels heterozygous, so that band is rewarded as present. There is no real heterozygous component outside the PAR; the homozygous (~cov) band is paralog / contamination in the body and non-discriminative backbone in the PAR, so it is treated as uninformative (like the frequent band).
|
◆ Parameters()
◆ print()
| void vg::Recombinator::Parameters::print |
( |
std::ostream & |
out | ) |
const |
Print a description of the parameters.
◆ absent_score
| double vg::Recombinator::Parameters::absent_score = ABSENT_SCORE |
Score for absent kmers. This should be less than 1 if we assume that having the right variants in the graph is more important than keeping the wrong variants out.
◆ badness_threshold
Badness threshold for subchains when using diploid sampling.
◆ banned_samples
| unordered_set<std::string> vg::Recombinator::Parameters::banned_samples |
Samples whose haplotypes shouldn't be used, even if they score well.
◆ buffer_size
| gbwt::size_type vg::Recombinator::Parameters::buffer_size = gbwt::DynamicGBWT::INSERT_BATCH_SIZE |
Buffer size (in nodes) for GBWT construction.
◆ coverage
| size_t vg::Recombinator::Parameters::coverage = COVERAGE |
Kmer coverage. Use 0 to estimate from kmer counts.
◆ diploid_sampling
| bool vg::Recombinator::Parameters::diploid_sampling = false |
After selecting the initial num_haplotypes haplotypes, choose the highest-scoring pair out of them.
◆ excluded_chains
| std::unordered_set<size_t> vg::Recombinator::Parameters::excluded_chains |
Top-level chains (by offset) to copy through verbatim instead of personalizing. The reference/generic path and all haplotypes in the chain are preserved.
◆ extra_fragments
| bool vg::Recombinator::Parameters::extra_fragments = false |
When using diploid sampling, include the remaining candidates as additional fragments in bad subchains.
◆ half_coverage_chains
| std::unordered_set<size_t> vg::Recombinator::Parameters::half_coverage_chains |
Top-level chains (by offset) to sample using the half-coverage model.
◆ half_coverage_num_haplotypes
| size_t vg::Recombinator::Parameters::half_coverage_num_haplotypes = 2 |
Number of haplotypes to generate for half-coverage chains.
◆ haploid_scoring
| bool vg::Recombinator::Parameters::haploid_scoring = false |
Use the haploid scoring model. The most common kmer count is used as the coverage estimate. Kmers that would be classified as heterozygous are treated as homozygous.
◆ het_adjustment
Additive term for adjusting the scores for heterozygous kmers after each haplotype to encourage even sampling of haplotypes with and without that kmer.
◆ high_coverage_chains
| std::unordered_set<size_t> vg::Recombinator::Parameters::high_coverage_chains |
Top-level chains (by offset) to sample using the high-coverage model.
◆ high_coverage_num_haplotypes
| size_t vg::Recombinator::Parameters::high_coverage_num_haplotypes = NUM_HAPLOTYPES |
Number of haplotypes to generate for high-coverage chains.
◆ include_reference
| bool vg::Recombinator::Parameters::include_reference = false |
Include named and reference paths.
◆ num_haplotypes
Number of haplotypes to be generated, or the number of candidates for diploid sampling.
◆ present_discount
Multiplicative factor for discounting the scores for present kmers after selecting a haplotype with that kmer.
◆ scoring_model
Scoring model for the chain currently being processed. Set per chain from high_coverage_chains / half_coverage_chains before scoring. Diploid sampling is not used with the non-standard models.
◆ wrap_contigs
| std::unordered_set<std::string> vg::Recombinator::Parameters::wrap_contigs |
Contig names whose origin (first) fragment of every generated haplotype should be doubled, creating a self-loop that wraps the end of the sequence back onto its start. Intended for circular contigs such as chrM. Matched against the top-level chain contig name.
The documentation for this struct was generated from the following files:
