AggregationType

class alphagenome.models.v1.variant_scorers.AggregationType(value)

Enum indicating the type of variant scorer aggregation.

DIFF_MEAN

Difference of means, i.e., mean(ALT) - mean(REF).

DIFF_SUM

Difference of sums, i.e., sum(ALT) - sum(REF).

DIFF_SUM_LOG2

Log scales predictions, then takes the sum and then the difference, i.e., sum(log2(ALT)) - sum(log2(REF)).

DIFF_LOG2_SUM

Takes the sum of predictions, applies a log transform, then takes the difference between predictions, i.e., log2(sum(ALT)) - log2(sum(REF)).

L2_DIFF

Takes the difference of ALT and REF predictions, then computes the L2 norm, i.e., l2_norm(ALT - REF).

L2_DIFF_LOG1P

Log scales the predictions + 1, takes the difference of ALT and REF predictions, then computes the L2 norm, i.e., l2_norm(log1p(ALT) - log1p(REF)).

ACTIVE_MEAN

Maximum of means, i.e., max(mean(ALT), mean(REF)).

ACTIVE_SUM

Maximum of sums, i.e., max(sum(ALT), sum(REF)).

to_proto()

Converts the aggregation type to its corresponding proto enum.