Rework compute_allele_frequency

python/tests/beagle_numba.py

@@ -708,7 +708,7 @@ def compute_individual_scores(alleles_1, allele_probs_1, alleles_2, allele_probs
     """
     n = len(alleles_1)  # Number of individuals
     assert len(alleles_2) == n, "Lengths of alleles differ."
-    assert n > 1, "There must be at least one individual."
+    assert n > 0, "There must be at least one individual."
     assert len(allele_probs_1) == n, "Lengths of alleles and probabilities differ."
     assert len(allele_probs_2) == n, "Lengths of alleles and probabilities differ."
     gt_probs = np.zeros((n, 3), dtype=np.float32)
@@ -780,7 +780,9 @@ def compute_dosage_r_squared(gt_probs):
     pass
 
 
-def compute_allele_frequency(gt_probs, *, allele=1):
+def compute_allele_frequency(
+    alleles_1, allele_probs_1, alleles_2, allele_probs_2, *, allele=1
+):
     """
     Estimate the frequency of an allele at a site from posterior genotype probabilities.
 
@@ -794,21 +796,25 @@ def compute_allele_frequency(gt_probs, *, allele=1):
     Am J Hum Genet. 84(2): 210–223. doi: 10.1016/j.ajhg.2009.01.005.
 
     In BEAGLE 4.1, AF: "Estimated ALT Allele Frequencies".
+    See `printInfo` in `VcfRecBuilder.java` of the BEAGLE 4.1 source code.
 
-    :param np.ndarray gt_probs: Genotype probabilities at a site.
+    :param numpy.ndarray alleles_1: Imputed alleles for haplotype 1.
+    :param numpy.ndarray allele_probs_1: Imputed allele probabilities for haplotype 1.
+    :param numpy.ndarray alleles_2: Imputed alleles for haplotype 2.
+    :param numpy.ndarray allele_probs_2: Imputed allele probabilities for haplotype 2.
     :param int allele: Allele (default = 1).
     :return: Estimated allele frequency.
     :rtype: float
     """
-    n = len(gt_probs)  # Number of individuals
+    n = len(alleles_1)  # Number of individuals
+    assert len(alleles_2) == n, "Lengths of alleles differ."
     assert n > 0, "There must be at least one individual."
+    assert len(allele_probs_1) == n, "Lengths of alleles and probabilities differ."
+    assert len(allele_probs_2) == n, "Lengths of alleles and probabilities differ."
     assert allele in [0, 1], f"Allele {allele} is not recognized."
-    if allele == 0:
-        est_allele_count_0 = 2 * np.sum(gt_probs[:, 0]) + np.sum(gt_probs[:, 1])
-        return est_allele_count_0 / (2 * n)
-    else:
-        est_allele_count_1 = np.sum(gt_probs[:, 1]) + 2 * np.sum(gt_probs[:, 2])
-        return est_allele_count_1 / (2 * n)
+    cum_allele_freq_1 = allele_probs_1[alleles_1 == allele]
+    cum_allele_freq_2 = allele_probs_2[alleles_2 == allele]
+    return (cum_allele_freq_1 + cum_allele_freq_2) / (2 * n)
 
 
 def write_vcf(impdata, out_file, *, chr_name="1"):