get_barcode_performance.py

def get_barcode_performance(csv_file, tree_file, lineage):
    meta = []
    for row in csv.DictReader(open(args.csv)):
        meta.append(row)

    try:
        t = ete3.Tree(args.tree,format=1)
    except:
        t = ete3.Tree(args.tree)

    nodes_performance = []
    for node in tqdm(t.traverse()):
        if node.is_leaf():
            continue
        leaves = set(node.get_leaf_names())
        training_tp = len([s for s in meta if s["type"]=="training" and s["id"] in leaves and s["lineage"]==lineage])
        if training_tp==0: continue

        training_fn = len([s for s in meta if s["type"]=="training" and s["id"] in leaves and s["lineage"]!=lineage])
        training_fp = len([s for s in meta if s["type"]=="training" and s["id"] not in leaves and s["lineage"]==lineage])
        training_tn = len([s for s in meta if s["type"]=="training" and s["id"] not in leaves and s["lineage"]!=lineage])

        training_sensitivity = training_tp/(training_tp+training_fn)
        training_specificity = training_tn/(training_tn+training_fp)
        if training_sensitivity>=0.95 and training_specificity>=0.95:
            test_tp = len([s for s in meta if s["type"]=="test" and s["id"] in leaves and s["lineage"]==lineage])
            test_fn = len([s for s in meta if s["type"]=="test" and s["id"] in leaves and s["lineage"]!=lineage])
            test_fp = len([s for s in meta if s["type"]=="test" and s["id"] not in leaves and s["lineage"]==lineage])
            test_tn = len([s for s in meta if s["type"]=="test" and s["id"] not in leaves and s["lineage"]!=lineage])

            test_sensitivity = test_tp/(test_tp+test_fn)
            test_specificity = test_tn/(test_tn+test_fp)
            nodes_performance.append((node.name,training_sensitivity,training_specificity, test_sensitivity,test_specificity))

    return tuple(sorted(nodes_performance, key=lambda x: (x[1],x[2]))[-1])

Hi :) Maybe something interesting for you https://codenpaste.com
would love to chat with you : drop me a line if you have time (using the link in the website for security reason). Thanks