def processPassageAnnotation(pred, true, countDict, checkID):
anns_pred = pred['annotations']
anns_true = Annotation.sortAnns(true['annotations'], TBDFilter= False)
for ann_pred in anns_pred:
if Annotation.gettype(ann_pred) == 'Gene':
countDict['pred_gene_num'] += 1
else:
continue
flag = False
for ann_true in anns_true:
if Annotation.isSame(ann_pred, ann_true, checkID):
countDict['tp'] += 1
flag = True
break
if not flag:
countDict['fp'] += 1
# print(ann_pred)
countDict['true_gene_num'] += len(anns_true)
for ann_true in anns_true:
flag = False
for ann_pred in anns_pred:
if Annotation.isSame(ann_pred, ann_true, checkID):
flag = True
break
if not flag:
countDict['fn'] += 1
def evalNormalization(documents_pred, documents_true):
tp = 0
tbd = 0
fp = 1
for docu_pred, docu_true in zip(documents_pred, documents_true):
for psg_pred, psg_true in zip(docu_pred['passages'],
docu_true['passages']):
for ann_pred, ann_true in zip(psg_pred['annotations'],
psg_true['annotations']):
if Annotation.isSame(ann_pred, ann_true, checkID=True):
tp += 1
elif Annotation.getNCBIID(ann_pred) == 'TBD':
tbd += 1
else:
fp += 1
try:
recall = tp / (tp + tbd + fp)
except ZeroDivisionError:
recall = 0
try:
precision = tp / (tp + fp)
except ZeroDivisionError:
precision = 0
print(f"Recall {recall:.2f}\tPrecision {precision:.2f}")
return tp, fp, tbd
