Predict fetal sex for a NIPTeR sample

Uses one or more NIPTeRSexModel objects to classify a sample as male or female. When multiple models are supplied, a majority vote determines the consensus call.

Usage

nipter_predict_sex(sample, ..., y_unique_ratio = NULL)

Arguments

sample

A NIPTeRSample object.

...

One or more NIPTeRSexModel objects, or a single list of models.

y_unique_ratio

Optional numeric scalar; a pre-computed Y-unique ratio (from nipter_y_unique_ratio) for the sample. This is only used when one of the models has method = "y_unique". If a "y_unique" model is present and this argument is NULL, that model is skipped with a warning.

Value

A list of class "NIPTeRSexPrediction" with elements:

prediction

Character; "male" or "female" (consensus when multiple models).

model_predictions

Named character vector of per-model predictions (names are the method names).

fractions

Named numeric vector with x_fraction and y_fraction for the sample.

sample_name

Character; the sample name.

Details

For each model, the sample's sex chromosome fractions are computed and classified using predict(model$model, newdata = ...). The classification is mapped to "male"/"female" using the model's male_cluster.

For "y_unique" models, the y_unique_ratio argument is used as the input feature instead of chromosome fractions derived from binned read counts.

With multiple models, the consensus is the label that appears most frequently (majority vote). In case of a tie, "female" is returned (conservative default for NIPT, as false-male calls could mask sex chromosome aneuploidies).

See also

nipter_sex_model(), nipter_sex_model_y_unique(), nipter_y_unique_ratio()

Examples

if (FALSE) { # \dontrun{
# Single model
pred <- nipter_predict_sex(test_sample, sex_model)
pred$prediction

# Consensus from three models (majority vote)
yr <- nipter_y_unique_ratio("sample.bam")
pred <- nipter_predict_sex(test_sample, model_y, model_xy, model_yu,
                           y_unique_ratio = yr$ratio)
pred$prediction
} # }