Mercurial > repos > goeckslab > image_learner

annotate plotly_plots.py @ 11:c5150cceab47 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
author goeckslab
date Sat, 18 Oct 2025 03:17:09 +0000
parents 85e6f4b2ad18
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
1 import json
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
2 from pathlib import Path
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
3 from typing import Dict, List, Optional
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
4
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
5 import numpy as np
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
6 import pandas as pd
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
7 import plotly.graph_objects as go
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
8 import plotly.io as pio
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
9 from constants import LABEL_COLUMN_NAME, SPLIT_COLUMN_NAME
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
10 from sklearn.metrics import auc, roc_curve
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
11 from sklearn.preprocessing import label_binarize
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
12
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
13
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
14 def build_classification_plots(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
15 test_stats_path: str,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
16 training_stats_path: Optional[str] = None,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
17 ) -> List[Dict[str, str]]:
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
18 """
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
19 Read Ludwig’s test_statistics.json and build three interactive Plotly panels:
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
20 - Confusion Matrix
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
21 - ROC-AUC
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
22 - Classification Report Heatmap
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
23
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
24 Returns a list of dicts, each with:
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
25 {
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
26 "title": <plot title>,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
27 "html": <HTML fragment for embedding>
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
28 }
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
29 """
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
30 # --- Load test stats ---
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
31 with open(test_stats_path, "r") as f:
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
32 test_stats = json.load(f)
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
33 label_stats = test_stats["label"]
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
34
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
35 # common sizing
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
36 cell = 40
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
37 n_classes = len(label_stats["confusion_matrix"])
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
38 side_px = max(cell * n_classes + 200, 600)
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
39 common_cfg = {"displayModeBar": True, "scrollZoom": True}
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
40
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
41 plots: List[Dict[str, str]] = []
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
42
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
43 # 0) Confusion Matrix
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
44 cm = np.array(label_stats["confusion_matrix"], dtype=int)
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
45 # Try to get actual class names from per_class_stats keys (which contain the real labels)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
46 pcs = label_stats.get("per_class_stats", {})
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
47 if pcs:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
48 labels = list(pcs.keys())
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
49 else:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
50 labels = label_stats.get("labels", [str(i) for i in range(cm.shape[0])])
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
51 total = cm.sum()
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
52
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
53 fig_cm = go.Figure(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
54 go.Heatmap(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
55 z=cm,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
56 x=labels,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
57 y=labels,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
58 colorscale="Blues",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
59 showscale=True,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
60 colorbar=dict(title="Count"),
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
61 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
62 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
63 fig_cm.update_traces(xgap=2, ygap=2)
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
64 fig_cm.update_layout(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
65 title=dict(text="Confusion Matrix", x=0.5),
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
66 xaxis_title="Predicted",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
67 yaxis_title="Observed",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
68 yaxis_autorange="reversed",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
69 width=side_px,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
70 height=side_px,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
71 margin=dict(t=100, l=80, r=80, b=80),
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
72 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
73
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
74 # annotate counts and percentages
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
75 mval = cm.max() if cm.size else 0
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
76 thresh = mval / 2
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
77 for i in range(cm.shape[0]):
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
78 for j in range(cm.shape[1]):
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
79 v = cm[i, j]
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
80 pct = (v / total * 100) if total > 0 else 0
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
81 color = "white" if v > thresh else "black"
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
82 fig_cm.add_annotation(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
83 x=labels[j],
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
84 y=labels[i],
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
85 text=f"<b>{v}</b>",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
86 showarrow=False,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
87 font=dict(color=color, size=14),
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
88 xanchor="center",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
89 yanchor="bottom",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
90 yshift=2,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
91 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
92 fig_cm.add_annotation(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
93 x=labels[j],
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
94 y=labels[i],
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
95 text=f"{pct:.1f}%",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
96 showarrow=False,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
97 font=dict(color=color, size=13),
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
98 xanchor="center",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
99 yanchor="top",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
100 yshift=-2,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
101 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
102
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
103 plots.append({
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
104 "title": "Confusion Matrix",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
105 "html": pio.to_html(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
106 fig_cm,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
107 full_html=False,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
108 include_plotlyjs="cdn",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
109 config=common_cfg
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
110 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
111 })
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
112
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
113 # 1) ROC-AUC Curves (Multi-class)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
114 roc_plot = _build_roc_auc_plot(test_stats_path, labels, common_cfg)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
115 if roc_plot:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
116 plots.append(roc_plot)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
117
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
118 # 2) Classification Report Heatmap
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
119 pcs = label_stats.get("per_class_stats", {})
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
120 if pcs:
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
121 classes = list(pcs.keys())
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
122 metrics = ["precision", "recall", "f1_score"]
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
123 z, txt = [], []
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
124 for c in classes:
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
125 row, trow = [], []
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
126 for m in metrics:
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
127 val = pcs[c].get(m, 0)
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
128 row.append(val)
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
129 trow.append(f"{val:.2f}")
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
130 z.append(row)
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
131 txt.append(trow)
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
132
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
133 fig_cr = go.Figure(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
134 go.Heatmap(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
135 z=z,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
136 x=metrics,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
137 y=[str(c) for c in classes],
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
138 text=txt,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
139 texttemplate="%{text}",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
140 colorscale="Reds",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
141 showscale=True,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
142 colorbar=dict(title="Value"),
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
143 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
144 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
145 fig_cr.update_layout(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
146 title="Classification Report",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
147 xaxis_title="",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
148 yaxis_title="Class",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
149 width=side_px,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
150 height=side_px,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
151 margin=dict(t=80, l=80, r=80, b=80),
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
152 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
153 plots.append({
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
154 "title": "Classification Report",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
155 "html": pio.to_html(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
156 fig_cr,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
157 full_html=False,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
158 include_plotlyjs=False,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
159 config=common_cfg
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
160 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
161 })
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
162
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
163 return plots
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
164
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
165
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
166 def _build_roc_auc_plot(test_stats_path: str, class_labels: List[str], config: dict) -> Optional[Dict[str, str]]:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
167 """
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
168 Build an interactive ROC-AUC curve plot for multi-class classification.
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
169 Following sklearn's ROC example with micro-average and per-class curves.
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
170
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
171 Args:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
172 test_stats_path: Path to test_statistics.json
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
173 class_labels: List of class label names
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
174 config: Plotly config dict
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
175
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
176 Returns:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
177 Dict with title and HTML, or None if data unavailable
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
178 """
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
179 try:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
180 # Get the experiment directory from test_stats_path
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
181 exp_dir = Path(test_stats_path).parent
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
182
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
183 # Load predictions with probabilities
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
184 predictions_path = exp_dir / "predictions.csv"
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
185 if not predictions_path.exists():
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
186 return None
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
187
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
188 df_pred = pd.read_csv(predictions_path)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
189
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
190 if SPLIT_COLUMN_NAME in df_pred.columns:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
191 split_series = df_pred[SPLIT_COLUMN_NAME].astype(str).str.lower()
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
192 test_mask = split_series.isin({"2", "test", "testing"})
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
193 if test_mask.any():
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
194 df_pred = df_pred[test_mask].reset_index(drop=True)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
195
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
196 if df_pred.empty:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
197 return None
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
198
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
199 # Extract probability columns (label_probabilities_0, label_probabilities_1, etc.)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
200 # or label_probabilities_<class_name> for string labels
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
201 prob_cols = [col for col in df_pred.columns if col.startswith('label_probabilities_') and col != 'label_probabilities']
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
202
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
203 # Sort by class number if numeric, otherwise keep alphabetical order
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
204 if prob_cols and prob_cols[0].split('_')[-1].isdigit():
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
205 prob_cols.sort(key=lambda x: int(x.split('_')[-1]))
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
206 else:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
207 prob_cols.sort() # Alphabetical sort for string class names
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
208
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
209 if not prob_cols:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
210 return None
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
211
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
212 # Get probabilities matrix (n_samples x n_classes)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
213 y_score = df_pred[prob_cols].values
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
214 n_classes = len(prob_cols)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
215
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
216 y_true = None
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
217 candidate_cols = [
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
218 LABEL_COLUMN_NAME,
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
219 f"{LABEL_COLUMN_NAME}_ground_truth",
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
220 f"{LABEL_COLUMN_NAME}__ground_truth",
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
221 f"{LABEL_COLUMN_NAME}_target",
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
222 f"{LABEL_COLUMN_NAME}__target",
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
223 ]
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
224 candidate_cols.extend(
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
225 [
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
226 col
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
227 for col in df_pred.columns
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
228 if (col.startswith(f"{LABEL_COLUMN_NAME}_") or col.startswith(f"{LABEL_COLUMN_NAME}__"))
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
229 and "probabilities" not in col
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
230 and "predictions" not in col
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
231 ]
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
232 )
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
233 for col in candidate_cols:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
234 if col in df_pred.columns and col not in prob_cols:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
235 y_true = df_pred[col].values
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
236 break
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
237
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
238 if y_true is None:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
239 desc_path = exp_dir / "description.json"
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
240 if desc_path.exists():
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
241 try:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
242 with open(desc_path, 'r') as f:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
243 desc = json.load(f)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
244 dataset_path = desc.get('dataset', '')
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
245 if dataset_path and Path(dataset_path).exists():
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
246 df_orig = pd.read_csv(dataset_path)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
247 if SPLIT_COLUMN_NAME in df_orig.columns:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
248 df_orig = df_orig[df_orig[SPLIT_COLUMN_NAME] == 2].reset_index(drop=True)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
249 if LABEL_COLUMN_NAME in df_orig.columns:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
250 y_true = df_orig[LABEL_COLUMN_NAME].values
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
251 if len(y_true) != len(df_pred):
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
252 print(
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
253 f"Warning: Test set size mismatch. Truncating to {len(df_pred)} samples for ROC plot."
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
254 )
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
255 y_true = y_true[:len(df_pred)]
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
256 else:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
257 print("Warning: Original dataset referenced in description.json is unavailable.")
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
258 except Exception as exc: # pragma: no cover - defensive
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
259 print(f"Warning: Failed to recover labels from dataset: {exc}")
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
260
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
261 if y_true is None or len(y_true) == 0:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
262 print("Warning: Unable to locate ground-truth labels for ROC plot.")
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
263 return None
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
264
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
265 if len(y_true) != len(y_score):
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
266 limit = min(len(y_true), len(y_score))
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
267 if limit == 0:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
268 return None
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
269 print(f"Warning: Aligning prediction and label lengths to {limit} samples for ROC plot.")
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
270 y_true = y_true[:limit]
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
271 y_score = y_score[:limit]
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
272
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
273 # Get actual class names from probability column names
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
274 actual_classes = [col.replace('label_probabilities_', '') for col in prob_cols]
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
275 display_classes = class_labels if len(class_labels) == n_classes else actual_classes
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
276
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
277 # Binarize the output following sklearn example
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
278 # Use actual class names if they're strings, otherwise use range
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
279 if isinstance(y_true[0], str):
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
280 y_test = label_binarize(y_true, classes=actual_classes)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
281 else:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
282 y_test = label_binarize(y_true, classes=list(range(n_classes)))
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
283
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
284 # Handle binary classification case
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
285 if y_test.ndim != 2:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
286 y_test = np.atleast_2d(y_test)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
287
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
288 if n_classes == 2:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
289 if y_test.shape[1] == 1:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
290 y_test = np.hstack([1 - y_test, y_test])
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
291 elif y_test.shape[1] != 2:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
292 print("Warning: Unexpected label binarization shape for binary ROC plot.")
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
293 return None
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
294 elif y_test.shape[1] != n_classes:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
295 print("Warning: Label binarization did not produce expected class dimension; skipping ROC plot.")
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
296 return None
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
297
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
298 # Compute ROC curve and ROC area for each class (following sklearn example)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
299 fpr = dict()
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
300 tpr = dict()
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
301 roc_auc = dict()
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
302
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
303 for i in range(n_classes):
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
304 if np.sum(y_test[:, i]) > 0: # Check if class exists in test set
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
305 fpr[i], tpr[i], _ = roc_curve(y_test[:, i], y_score[:, i])
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
306 roc_auc[i] = auc(fpr[i], tpr[i])
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
307
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
308 # Compute micro-average ROC curve and ROC area (sklearn example)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
309 fpr["micro"], tpr["micro"], _ = roc_curve(y_test.ravel(), y_score.ravel())
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
310 roc_auc["micro"] = auc(fpr["micro"], tpr["micro"])
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
311
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
312 # Create ROC curve plot
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
313 fig_roc = go.Figure()
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
314
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
315 # Colors for different classes
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
316 colors = [
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
317 '#1f77b4', '#ff7f0e', '#2ca02c', '#d62728', '#9467bd',
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
318 '#8c564b', '#e377c2', '#7f7f7f', '#bcbd22', '#17becf'
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
319 ]
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
320
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
321 # Plot micro-average ROC curve first (most important)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
322 fig_roc.add_trace(go.Scatter(
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
323 x=fpr["micro"],
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
324 y=tpr["micro"],
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
325 mode='lines',
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
326 name=f'Micro-average ROC (AUC = {roc_auc["micro"]:.3f})',
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
327 line=dict(color='deeppink', width=3, dash='dot'),
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
328 hovertemplate=('<b>Micro-average ROC</b><br>'
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
329 'FPR: %{x:.3f}<br>'
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
330 'TPR: %{y:.3f}<br>'
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
331 f'AUC: {roc_auc["micro"]:.3f}<extra></extra>')
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
332 ))
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
333
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
334 # Plot ROC curve for each class
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
335 for i in range(n_classes):
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
336 if i in roc_auc: # Only plot if class exists in test set
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
337 class_name = display_classes[i] if i < len(display_classes) else f"Class {i}"
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
338 color = colors[i % len(colors)]
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
339
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
340 fig_roc.add_trace(go.Scatter(
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
341 x=fpr[i],
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
342 y=tpr[i],
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
343 mode='lines',
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
344 name=f'{class_name} (AUC = {roc_auc[i]:.3f})',
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
345 line=dict(color=color, width=2),
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
346 hovertemplate=(f'<b>{class_name}</b><br>'
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
347 'FPR: %{x:.3f}<br>'
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
348 'TPR: %{y:.3f}<br>'
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
349 f'AUC: {roc_auc[i]:.3f}<extra></extra>')
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
350 ))
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
351
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
352 # Add diagonal line (random classifier)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
353 fig_roc.add_trace(go.Scatter(
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
354 x=[0, 1],
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
355 y=[0, 1],
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
356 mode='lines',
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
357 name='Random Classifier',
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
358 line=dict(color='gray', width=1, dash='dash'),
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
359 hovertemplate='Random Classifier<br>AUC = 0.500<extra></extra>'
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
360 ))
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
361
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
362 # Calculate macro-average AUC
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
363 class_aucs = [roc_auc[i] for i in range(n_classes) if i in roc_auc]
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
364 if class_aucs:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
365 macro_auc = np.mean(class_aucs)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
366 title_text = f"ROC Curves (Micro-avg = {roc_auc['micro']:.3f}, Macro-avg = {macro_auc:.3f})"
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
367 else:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
368 title_text = f"ROC Curves (Micro-avg = {roc_auc['micro']:.3f})"
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
369
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
370 fig_roc.update_layout(
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
371 title=dict(text=title_text, x=0.5),
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
372 xaxis_title="False Positive Rate",
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
373 yaxis_title="True Positive Rate",
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
374 width=700,
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
375 height=600,
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
376 margin=dict(t=80, l=80, r=80, b=80),
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
377 legend=dict(
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
378 x=0.6,
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
379 y=0.1,
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
380 bgcolor="rgba(255,255,255,0.9)",
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
381 bordercolor="rgba(0,0,0,0.2)",
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
382 borderwidth=1
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
383 ),
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
384 hovermode='closest'
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
385 )
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
386
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
387 # Set equal aspect ratio and proper range
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
388 fig_roc.update_xaxes(range=[0, 1.0])
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
389 fig_roc.update_yaxes(range=[0, 1.05])
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
390
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
391 return {
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
392 "title": "ROC-AUC Curves",
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
393 "html": pio.to_html(
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
394 fig_roc,
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
395 full_html=False,
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
396 include_plotlyjs=False,
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
397 config=config
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
398 )
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
399 }
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
400
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
401 except Exception as e:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
402 print(f"Error building ROC-AUC plot: {e}")
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
403 return None