def test_unique_clusters(self):
protoclusters = [create_protocluster(i, 10, product=prod) for i, prod in enumerate("ABC")]
candidates = [CandidateCluster(CandidateCluster.kinds.INTERLEAVED, protoclusters[:2]),
CandidateCluster(CandidateCluster.kinds.INTERLEAVED, protoclusters[1:])]
assert protoclusters[1] in candidates[0].protoclusters and protoclusters[1] in candidates[1].protoclusters
region = Region(candidate_clusters=candidates)
unique_clusters = region.get_unique_protoclusters()
# if the protocluster in both candidates is repeated, there'll be an extra
assert len(unique_clusters) == 3
assert unique_clusters == protoclusters
def score_against_protoclusters(label: str, region: Region,
hits_by_reference: HitsByReference,
query_components: Dict[CDSCollection,
Components],
mode: Mode) -> VariantResults:
""" Performs a protocluster vs protocluster comparison
Arguments:
label: the name to attach to the results
region: the query Region
hits_by_reference: a dictionary mapping ReferenceRecord to
a dictionary mapping reference CDS name to Hit
query_components: a dictionary mapping the region and each contained protocluster to
a Components instance with the relevant data
mode: the Mode in which to run the analysis
Returns:
a VariantResults instance
"""
score_matrix: Dict[int, Dict[ReferenceRegion, Dict[
ReferenceProtocluster,
ReferenceScorer]]] = defaultdict(lambda: defaultdict(dict))
reference_best_scores: Dict[Protocluster, Dict[
ReferenceRegion, float]] = defaultdict(lambda: defaultdict(float))
local_hits = filter_by_query_area(region, hits_by_reference)
for ref_region in local_hits:
hits_for_ref_region = {ref_region: local_hits[ref_region]}
for ref_protocluster in ref_region.protoclusters:
hits = filter_by_reference_protocluster(ref_protocluster,
hits_for_ref_region)
for protocluster in region.get_unique_protoclusters():
for scorer in score_query_area(protocluster, hits,
query_components[protocluster],
mode):
score = max(
scorer.final_score,
reference_best_scores[protocluster][ref_region])
reference_best_scores[protocluster][ref_region] = score
score_matrix[protocluster.get_protocluster_number(
)][ref_region][ref_protocluster] = scorer
reference_total_scores: Dict[ReferenceRegion, float] = defaultdict(float)
for ref_region_to_score in reference_best_scores.values():
for ref_region, score in ref_region_to_score.items():
reference_total_scores[ref_region] += score
region_ranking = sorted(reference_total_scores.items(),
key=lambda x: x[1],
reverse=True)
region_ranking, score_matrix, best_hits = apply_limits_to_rankings(
region_ranking, score_matrix, local_hits)
return VariantResults(label, region_ranking,
ProtoToProtoScores(score_matrix), best_hits)
def score_as_protoclusters(label: str, region: Region,
hits_by_reference: HitsByReference,
query_components: Dict[CDSCollection, Components],
mode: Mode) -> VariantResults:
""" Performs a protocluster vs reference region comparison
Arguments:
label: the name to attach to the results
region: the query Region
hits_by_reference: a dictionary mapping ReferenceRecord to
a dictionary mapping reference CDS name to Hit
query_components: a dictionary mapping the region and each contained protocluster to
a Components instance with the relevant data
mode: the Mode in which to run the analysis
Returns:
a VariantResults instance
"""
local_hits = filter_by_query_area(region, hits_by_reference)
total_scores: Dict[ReferenceRegion, float] = defaultdict(float)
scores: Dict[int, Dict[ReferenceRegion,
ReferenceScorer]] = defaultdict(dict)
for protocluster in region.get_unique_protoclusters():
for scorer in score_query_area(protocluster, local_hits,
query_components[protocluster], mode):
total_scores[scorer.reference] += calculate_protocluster_ranking(
scorer)
scores[protocluster.get_protocluster_number()][
scorer.reference] = scorer
ranking = sorted(total_scores.items(), key=lambda x: x[1], reverse=True)
ranking, scores, best_hits = apply_limits_to_rankings(
ranking, scores, local_hits)
return VariantResults(label, ranking, ProtoToRegionScores(scores),
best_hits)
