infer

The infer command identifies candidate introgressed tracts from genotype data.

It reads preprocessing settings from the sstar2 configuration file, computes features for genomic windows, applies a trained ONNX model, and writes windows whose observed S*_score is greater than the predicted score to a BED file.

Example

Using the trained model from the train command and the configuration file, users can infer the introgressed fragments with the following command:

sstar2 infer --model examples/results/sstar2.example.trained.model.onnx \
             --config examples/data/sstar2.example.config.yaml \
             --feat-file examples/results/sstar2.example.inference.features.tsv \
             --pred-file examples/results/sstar2.example.pred.tsv \
             --tract-file examples/results/sstar2.example.inferred.tracts.bed

The inferred tracts can be found here. Two additional files are generated: one records the features for prediction, and the other contains the observed and predicted S* scores.

Outputs

• --feat-file: feature TSV generated from the input genotype data specified in the configuration file.
• --pred-file: prediction TSV containing the observed and predicted S* scores.
• --tract-file: BED file of inferred tracts.

The BED file is tab-separated and has no header:

chrom  start  end  sample

The start coordinate is written in BED-style 0-based format.

Settings

Argument	Description
--model	Path to the trained model file.
--config	Path to the sstar2 configuration YAML file.
--feat-file	Path to the feature TSV file.
--pred-file	Path to the prediction TSV file.
--tract-file	Path to the output BED file.
--match-bonus	Bonus for matching genotypes between two variants. Default: 5000.
--max-mismatch	Maximum genotype distance allowed before a pair is discarded. Default: 5.
--mismatch-penalty	Penalty for mismatching genotypes between two variants. Default: -10000.