abd-clam 0.11.0

Clustered Learning of Approximate Manifolds
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148

import logging
import pathlib
import re
import time
import typing

import csv_space
import datatable
import numpy
from sklearn.metrics import accuracy_score
from sklearn.model_selection import train_test_split
from utils import nyxus_metadata

from abd_clam import metric
from abd_clam.classification import classifier
from abd_clam.core import cluster_criteria

logging.basicConfig(
    format="%(asctime)s - %(levelname)s - %(name)s - %(funcName)s - %(message)s",
    datefmt="%d-%b-%y %H:%M:%S",
)


class NyxusDataset(csv_space.CsvDataset):
    def __init__(
        self,
        path: pathlib.Path,
        name: str,
        *,
        file_names_column: str,
        class_labels: typing.Literal["platform", "tissue"],
        non_feature_columns: list[str] = None,
        indices: list[int] = None,
        normalize: bool = True,
        means: numpy.ndarray = None,
        sds: numpy.ndarray = None,
    ) -> None:
        data: datatable.Frame = datatable.fread(str(path))

        assert (
            file_names_column in data.names
        ), f"Column {file_names_column} not found in the dataset."

        file_names: list[str] = data[file_names_column].to_list()[0]
        pattern = re.compile(
            "(?P<d>\\w+)_p(?P<p>\\d+)_y(?P<y>\\d+)_r(?P<r>\\d+)_c(?P<c>\\d+).tif",
        )
        matches = list(map(pattern.match, file_names))

        if class_labels == "tissue":
            labels = list(map(int, (r.group("y") for r in matches)))
        else:
            labels = list(map(int, (r.group("p") for r in matches)))

        super().__init__(
            path=path,
            name=name,
            full_data=data,
            labels=labels,
            non_feature_columns=non_feature_columns,
            indices=indices,
            normalize=normalize,
            means=means,
            sds=sds,
        )


def read_nyxus_cells():
    return NyxusDataset(
        path=nyxus_metadata.CELLS_CSV_PATH,
        name="nyxus_cells",
        file_names_column="mask_image",
        class_labels="tissue",
        non_feature_columns=nyxus_metadata.CELLS_NON_FEATURES,
    )


def read_nyxus_images():
    return NyxusDataset(
        path=nyxus_metadata.IMAGES_CSV_PATH,
        name="nyxus_images",
        file_names_column="mask_image",
        class_labels="tissue",
        non_feature_columns=nyxus_metadata.IMAGES_NON_FEATURES,
    )


# noinspection DuplicatedCode
def experiment():
    nyxus_full = read_nyxus_images()

    test_size = 0.2 if nyxus_full.cardinality <= 50_000 else 10_000
    train_indices, test_indices, train_labels, test_labels = train_test_split(
        nyxus_full.indices,
        nyxus_full.labels,
        test_size=test_size,
        random_state=42,
        stratify=nyxus_full.labels,
    )

    nyxus_train = nyxus_full.subset(
        train_indices,
        f"{nyxus_full.name}_train",
        labels=train_labels,
    )
    nyxus_test = nyxus_full.subset(
        test_indices,
        f"{nyxus_full.name}_test",
        labels=test_labels,
    )

    nyxus_spaces = [
        csv_space.CsvSpace(nyxus_train, metric.ScipyMetric("euclidean"), False),
        csv_space.CsvSpace(nyxus_train, metric.ScipyMetric("cityblock"), False),
    ]

    start = time.perf_counter()
    nyxus_classifier = classifier.Classifier(
        nyxus_train.labels,
        nyxus_spaces,
        partition_criteria=[cluster_criteria.MaxDepth(20)],
    ).build()
    end = time.perf_counter()
    build_time = end - start

    start = time.perf_counter()
    predicted_labels, _ = nyxus_classifier.predict(nyxus_test)
    end = time.perf_counter()
    prediction_time = end - start

    score = accuracy_score(nyxus_test.labels, predicted_labels)

    # 21:47:03  22:36:14

    print(f"Building the {nyxus_full.name} classifier for:")
    print(f"\t{nyxus_train.cardinality} instances and")
    print(f"\t{nyxus_classifier.unique_labels} unique labels")
    print(f"\ttook {build_time:.2e} seconds.")

    print("Predicting from the classifier for:")
    print(f"\t{nyxus_test.cardinality} instances took")
    print(f"\ttook {prediction_time:.2e} seconds.")

    print(f"The accuracy score was {score:.3f}")


if __name__ == "__main__":
    experiment()