iaode 0.2.9__tar.gz → 0.3.1__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {iaode-0.2.9/iaode.egg-info → iaode-0.3.1}/PKG-INFO +72 -30
- {iaode-0.2.9 → iaode-0.3.1}/README.md +70 -29
- {iaode-0.2.9 → iaode-0.3.1}/iaode/__init__.py +1 -1
- {iaode-0.2.9 → iaode-0.3.1}/iaode/agent.py +4 -2
- {iaode-0.2.9 → iaode-0.3.1/iaode.egg-info}/PKG-INFO +72 -30
- {iaode-0.2.9 → iaode-0.3.1}/iaode.egg-info/SOURCES.txt +1 -0
- {iaode-0.2.9 → iaode-0.3.1}/iaode.egg-info/requires.txt +1 -0
- {iaode-0.2.9 → iaode-0.3.1}/pyproject.toml +2 -1
- iaode-0.3.1/tests/test_pages_metadata.py +11 -0
- {iaode-0.2.9 → iaode-0.3.1}/LICENSE +0 -0
- {iaode-0.2.9 → iaode-0.3.1}/MANIFEST.in +0 -0
- {iaode-0.2.9 → iaode-0.3.1}/iaode/BEN.py +0 -0
- {iaode-0.2.9 → iaode-0.3.1}/iaode/DRE.py +0 -0
- {iaode-0.2.9 → iaode-0.3.1}/iaode/LSE.py +0 -0
- {iaode-0.2.9 → iaode-0.3.1}/iaode/annotation.py +0 -0
- {iaode-0.2.9 → iaode-0.3.1}/iaode/datasets.py +0 -0
- {iaode-0.2.9 → iaode-0.3.1}/iaode/environment.py +0 -0
- {iaode-0.2.9 → iaode-0.3.1}/iaode/mixin.py +0 -0
- {iaode-0.2.9 → iaode-0.3.1}/iaode/model.py +0 -0
- {iaode-0.2.9 → iaode-0.3.1}/iaode/module.py +0 -0
- {iaode-0.2.9 → iaode-0.3.1}/iaode/utils.py +0 -0
- {iaode-0.2.9 → iaode-0.3.1}/iaode.egg-info/dependency_links.txt +0 -0
- {iaode-0.2.9 → iaode-0.3.1}/iaode.egg-info/top_level.txt +0 -0
- {iaode-0.2.9 → iaode-0.3.1}/requirements.txt +0 -0
- {iaode-0.2.9 → iaode-0.3.1}/setup.cfg +0 -0
- {iaode-0.2.9 → iaode-0.3.1}/setup.py +0 -0
- {iaode-0.2.9 → iaode-0.3.1}/tests/test_smoke.py +0 -0
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.4
|
|
2
2
|
Name: iaode
|
|
3
|
-
Version: 0.
|
|
3
|
+
Version: 0.3.1
|
|
4
4
|
Summary: Interpretable Autoencoder with Ordinary Differential Equations for single-cell omics analysis
|
|
5
5
|
Author-email: Zeyu Fu <fuzeyu99@126.com>
|
|
6
6
|
License-Expression: MIT
|
|
@@ -31,6 +31,7 @@ Requires-Dist: torchdiffeq>=0.2.0
|
|
|
31
31
|
Requires-Dist: anndata>=0.8.0
|
|
32
32
|
Requires-Dist: scanpy>=1.8.0
|
|
33
33
|
Requires-Dist: scvi-tools>=0.16.0
|
|
34
|
+
Requires-Dist: zarr>=2.0.0
|
|
34
35
|
Requires-Dist: requests>=2.25.0
|
|
35
36
|
Requires-Dist: tqdm>=4.60.0
|
|
36
37
|
Provides-Extra: dev
|
|
@@ -47,18 +48,36 @@ Requires-Dist: sphinx-rtd-theme>=1.0.0; extra == "docs"
|
|
|
47
48
|
Requires-Dist: sphinx-autodoc-typehints>=1.18.0; extra == "docs"
|
|
48
49
|
Dynamic: license-file
|
|
49
50
|
|
|
51
|
+
<div align="center">
|
|
52
|
+
<a href="https://peterponyu.github.io/">
|
|
53
|
+
<img src="https://peterponyu.github.io/assets/badges/iAODE.svg" width="64" alt="ZF Lab · iAODE">
|
|
54
|
+
</a>
|
|
55
|
+
</div>
|
|
56
|
+
|
|
50
57
|
# iAODE: Interpretable Accessibility ODE VAE for scATAC-seq
|
|
51
58
|
|
|
52
|
-
[](https://pypi.org/project/iaode/) [](https://www.python.org/downloads/) [](https://opensource.org/licenses/MIT) [](https://pytorch.org/) [](https://github.com/PeterPonyu/iAODE/actions/workflows/test.yml)
|
|
59
|
+
[](https://pypi.org/project/iaode/) [](https://www.python.org/downloads/) [](https://opensource.org/licenses/MIT) [](https://pytorch.org/) [](https://github.com/PeterPonyu/iAODE/actions/workflows/test.yml) [](https://doi.org/10.5281/zenodo.18453104)
|
|
60
|
+
|
|
61
|
+
> **Interactive Tools:** The dataset browser and continuity explorer have moved to [**SCPortal**](https://peterponyu.github.io/scportal/) — a unified single-cell data portal with expanded datasets, LAIOR benchmarks, and more. The [original project page](https://peterponyu.github.io/iAODE/) is preserved as the manuscript publication snapshot.
|
|
62
|
+
|
|
63
|
+
## Public Pages and local application surfaces
|
|
64
|
+
|
|
65
|
+
`site/` is the public GitHub Pages landing at the canonical URL
|
|
66
|
+
`https://peterponyu.github.io/iAODE/`. It provides project documentation and
|
|
67
|
+
reproducible entry points only. No `/iAODE/frontend/` Pages route is supported.
|
|
68
|
+
|
|
69
|
+
`frontend/` and `api/static/` are local application assets for the training UI.
|
|
70
|
+
They remain local so model training and data handling are performed in the
|
|
71
|
+
user's environment rather than on the public Pages site.
|
|
53
72
|
|
|
54
|
-
**iAODE** (Interpretable Accessibility ODE VAE) is a lightweight deep learning framework
|
|
73
|
+
**iAODE** (Interpretable Accessibility ODE VAE) is a lightweight deep learning framework centered on single-cell ATAC-seq (scATAC-seq) data. It integrates a Variational Autoencoder (VAE) with a Neural ODE and an interpretable bottleneck to support:
|
|
55
74
|
|
|
56
|
-
1. **
|
|
75
|
+
1. **Count-based Modeling:** Handles sparse accessibility profiles with Negative Binomial (NB) or Zero-Inflated Negative Binomial (ZINB) likelihoods.
|
|
57
76
|
2. **Continuous Trajectory Inference:** Infers dynamics via Neural ODE pseudotime and latent velocity fields.
|
|
58
77
|
3. **Interpretable Latent Factors:** Utilizes a biologically aligned bottleneck to extract meaningful features.
|
|
59
78
|
4. **Scalable Preprocessing:** Implements TF‑IDF normalization and Highly Variable Peak (HVP) selection, aligned with Signac and SnapATAC2 best practices.
|
|
60
79
|
|
|
61
|
-
The
|
|
80
|
+
The core package targets scATAC-seq data, including sparse peak counts, library size variability, heterogeneous accessibility kinetics, and dynamic regulatory trajectories. Some bundled scripts and UI workflows also cover scRNA-seq comparison tasks, but scATAC-seq remains the primary scope of the repository.
|
|
62
81
|
|
|
63
82
|
---
|
|
64
83
|
|
|
@@ -98,39 +117,58 @@ Loss = recon + i_recon + ODE_consistency + β·KL + dip·DIP + tc·TC + info·MM
|
|
|
98
117
|
|
|
99
118
|
## Installation
|
|
100
119
|
|
|
101
|
-
###
|
|
102
|
-
|
|
120
|
+
### From PyPI (Recommended)
|
|
121
|
+
|
|
103
122
|
```bash
|
|
104
123
|
pip install iaode
|
|
105
124
|
```
|
|
106
125
|
|
|
107
|
-
###
|
|
108
|
-
For users who want the **Interactive Training UI**, you must clone the repository (since the UI requires the server files):
|
|
126
|
+
### From Source
|
|
109
127
|
|
|
110
128
|
```bash
|
|
111
129
|
git clone https://github.com/PeterPonyu/iAODE.git
|
|
112
130
|
cd iAODE
|
|
113
131
|
pip install -e .
|
|
114
|
-
pip install fastapi uvicorn # Required for the UI server
|
|
115
132
|
```
|
|
116
133
|
|
|
134
|
+
### Requirements
|
|
135
|
+
|
|
136
|
+
- Python ≥ 3.9
|
|
137
|
+
- PyTorch ≥ 1.10.0
|
|
138
|
+
- AnnData ≥ 0.8.0
|
|
139
|
+
- Scanpy ≥ 1.8.0
|
|
140
|
+
- scvi-tools ≥ 0.16.0
|
|
141
|
+
|
|
142
|
+
See `requirements.txt` for the complete list of dependencies.
|
|
143
|
+
|
|
117
144
|
---
|
|
145
|
+
|
|
118
146
|
## Quick Start
|
|
119
147
|
|
|
120
|
-
###
|
|
121
|
-
**No Node.js required.** The interface is pre-built and served via Python.
|
|
148
|
+
### Training UI (Web Interface)
|
|
122
149
|
|
|
123
|
-
|
|
124
|
-
|
|
125
|
-
|
|
126
|
-
|
|
127
|
-
|
|
128
|
-
|
|
129
|
-
|
|
130
|
-
|
|
131
|
-
|
|
132
|
-
|
|
133
|
-
|
|
150
|
+
iAODE includes an integrated web-based training interface for interactive model training:
|
|
151
|
+
|
|
152
|
+
```bash
|
|
153
|
+
# Start the training UI (Python - cross-platform)
|
|
154
|
+
python start_training_ui.py
|
|
155
|
+
|
|
156
|
+
# Or use the shell script (Linux/macOS)
|
|
157
|
+
./start_training_ui.sh
|
|
158
|
+
```
|
|
159
|
+
|
|
160
|
+
Then open your browser to:
|
|
161
|
+
- **Training UI**: http://localhost:8000/ui
|
|
162
|
+
- **API Documentation**: http://localhost:8000/docs
|
|
163
|
+
|
|
164
|
+
**Features:**
|
|
165
|
+
- Upload scATAC-seq data and related `.h5ad` workflows through the UI
|
|
166
|
+
- Configure preprocessing parameters (TF-IDF, HVP selection, subsampling)
|
|
167
|
+
- Set training parameters interactively
|
|
168
|
+
- Monitor training progress in real-time
|
|
169
|
+
- Download trained embeddings (latent and interpretable)
|
|
170
|
+
|
|
171
|
+
---
|
|
134
172
|
|
|
135
173
|
### Basic scATAC-seq Workflow
|
|
136
174
|
|
|
@@ -317,7 +355,7 @@ iaode.datasets.clear_cache() # Clear all cached data
|
|
|
317
355
|
|
|
318
356
|
## Examples
|
|
319
357
|
|
|
320
|
-
|
|
358
|
+
Examples are provided in the `examples/` directory. See **[examples/README.md](examples/README.md)** for details.
|
|
321
359
|
|
|
322
360
|
### Example Scripts
|
|
323
361
|
|
|
@@ -604,12 +642,16 @@ for epoch in range(100):
|
|
|
604
642
|
If you use iAODE in your research, please cite:
|
|
605
643
|
|
|
606
644
|
```bibtex
|
|
607
|
-
@
|
|
608
|
-
|
|
609
|
-
|
|
610
|
-
|
|
611
|
-
|
|
612
|
-
|
|
645
|
+
@article{Fu2026iAODE,
|
|
646
|
+
author = {Fu, Zeyu and Chen, Chunlin and Wang, Song and Wang, Junping and Chen, Shilei},
|
|
647
|
+
title = {iAODE for benchmarking and continuum modeling of single-cell chromatin accessibility},
|
|
648
|
+
journal = {Communications Biology},
|
|
649
|
+
volume = {9},
|
|
650
|
+
number = {1},
|
|
651
|
+
pages = {507},
|
|
652
|
+
year = {2026},
|
|
653
|
+
doi = {10.1038/s42003-026-09768-8},
|
|
654
|
+
url = {https://doi.org/10.1038/s42003-026-09768-8}
|
|
613
655
|
}
|
|
614
656
|
```
|
|
615
657
|
|
|
@@ -1,15 +1,33 @@
|
|
|
1
|
+
<div align="center">
|
|
2
|
+
<a href="https://peterponyu.github.io/">
|
|
3
|
+
<img src="https://peterponyu.github.io/assets/badges/iAODE.svg" width="64" alt="ZF Lab · iAODE">
|
|
4
|
+
</a>
|
|
5
|
+
</div>
|
|
6
|
+
|
|
1
7
|
# iAODE: Interpretable Accessibility ODE VAE for scATAC-seq
|
|
2
8
|
|
|
3
|
-
[](https://pypi.org/project/iaode/) [](https://www.python.org/downloads/) [](https://opensource.org/licenses/MIT) [](https://pytorch.org/) [](https://github.com/PeterPonyu/iAODE/actions/workflows/test.yml)
|
|
9
|
+
[](https://pypi.org/project/iaode/) [](https://www.python.org/downloads/) [](https://opensource.org/licenses/MIT) [](https://pytorch.org/) [](https://github.com/PeterPonyu/iAODE/actions/workflows/test.yml) [](https://doi.org/10.5281/zenodo.18453104)
|
|
10
|
+
|
|
11
|
+
> **Interactive Tools:** The dataset browser and continuity explorer have moved to [**SCPortal**](https://peterponyu.github.io/scportal/) — a unified single-cell data portal with expanded datasets, LAIOR benchmarks, and more. The [original project page](https://peterponyu.github.io/iAODE/) is preserved as the manuscript publication snapshot.
|
|
12
|
+
|
|
13
|
+
## Public Pages and local application surfaces
|
|
14
|
+
|
|
15
|
+
`site/` is the public GitHub Pages landing at the canonical URL
|
|
16
|
+
`https://peterponyu.github.io/iAODE/`. It provides project documentation and
|
|
17
|
+
reproducible entry points only. No `/iAODE/frontend/` Pages route is supported.
|
|
18
|
+
|
|
19
|
+
`frontend/` and `api/static/` are local application assets for the training UI.
|
|
20
|
+
They remain local so model training and data handling are performed in the
|
|
21
|
+
user's environment rather than on the public Pages site.
|
|
4
22
|
|
|
5
|
-
**iAODE** (Interpretable Accessibility ODE VAE) is a lightweight deep learning framework
|
|
23
|
+
**iAODE** (Interpretable Accessibility ODE VAE) is a lightweight deep learning framework centered on single-cell ATAC-seq (scATAC-seq) data. It integrates a Variational Autoencoder (VAE) with a Neural ODE and an interpretable bottleneck to support:
|
|
6
24
|
|
|
7
|
-
1. **
|
|
25
|
+
1. **Count-based Modeling:** Handles sparse accessibility profiles with Negative Binomial (NB) or Zero-Inflated Negative Binomial (ZINB) likelihoods.
|
|
8
26
|
2. **Continuous Trajectory Inference:** Infers dynamics via Neural ODE pseudotime and latent velocity fields.
|
|
9
27
|
3. **Interpretable Latent Factors:** Utilizes a biologically aligned bottleneck to extract meaningful features.
|
|
10
28
|
4. **Scalable Preprocessing:** Implements TF‑IDF normalization and Highly Variable Peak (HVP) selection, aligned with Signac and SnapATAC2 best practices.
|
|
11
29
|
|
|
12
|
-
The
|
|
30
|
+
The core package targets scATAC-seq data, including sparse peak counts, library size variability, heterogeneous accessibility kinetics, and dynamic regulatory trajectories. Some bundled scripts and UI workflows also cover scRNA-seq comparison tasks, but scATAC-seq remains the primary scope of the repository.
|
|
13
31
|
|
|
14
32
|
---
|
|
15
33
|
|
|
@@ -49,39 +67,58 @@ Loss = recon + i_recon + ODE_consistency + β·KL + dip·DIP + tc·TC + info·MM
|
|
|
49
67
|
|
|
50
68
|
## Installation
|
|
51
69
|
|
|
52
|
-
###
|
|
53
|
-
|
|
70
|
+
### From PyPI (Recommended)
|
|
71
|
+
|
|
54
72
|
```bash
|
|
55
73
|
pip install iaode
|
|
56
74
|
```
|
|
57
75
|
|
|
58
|
-
###
|
|
59
|
-
For users who want the **Interactive Training UI**, you must clone the repository (since the UI requires the server files):
|
|
76
|
+
### From Source
|
|
60
77
|
|
|
61
78
|
```bash
|
|
62
79
|
git clone https://github.com/PeterPonyu/iAODE.git
|
|
63
80
|
cd iAODE
|
|
64
81
|
pip install -e .
|
|
65
|
-
pip install fastapi uvicorn # Required for the UI server
|
|
66
82
|
```
|
|
67
83
|
|
|
84
|
+
### Requirements
|
|
85
|
+
|
|
86
|
+
- Python ≥ 3.9
|
|
87
|
+
- PyTorch ≥ 1.10.0
|
|
88
|
+
- AnnData ≥ 0.8.0
|
|
89
|
+
- Scanpy ≥ 1.8.0
|
|
90
|
+
- scvi-tools ≥ 0.16.0
|
|
91
|
+
|
|
92
|
+
See `requirements.txt` for the complete list of dependencies.
|
|
93
|
+
|
|
68
94
|
---
|
|
95
|
+
|
|
69
96
|
## Quick Start
|
|
70
97
|
|
|
71
|
-
###
|
|
72
|
-
**No Node.js required.** The interface is pre-built and served via Python.
|
|
98
|
+
### Training UI (Web Interface)
|
|
73
99
|
|
|
74
|
-
|
|
75
|
-
|
|
76
|
-
|
|
77
|
-
|
|
78
|
-
|
|
79
|
-
|
|
80
|
-
|
|
81
|
-
|
|
82
|
-
|
|
83
|
-
|
|
84
|
-
|
|
100
|
+
iAODE includes an integrated web-based training interface for interactive model training:
|
|
101
|
+
|
|
102
|
+
```bash
|
|
103
|
+
# Start the training UI (Python - cross-platform)
|
|
104
|
+
python start_training_ui.py
|
|
105
|
+
|
|
106
|
+
# Or use the shell script (Linux/macOS)
|
|
107
|
+
./start_training_ui.sh
|
|
108
|
+
```
|
|
109
|
+
|
|
110
|
+
Then open your browser to:
|
|
111
|
+
- **Training UI**: http://localhost:8000/ui
|
|
112
|
+
- **API Documentation**: http://localhost:8000/docs
|
|
113
|
+
|
|
114
|
+
**Features:**
|
|
115
|
+
- Upload scATAC-seq data and related `.h5ad` workflows through the UI
|
|
116
|
+
- Configure preprocessing parameters (TF-IDF, HVP selection, subsampling)
|
|
117
|
+
- Set training parameters interactively
|
|
118
|
+
- Monitor training progress in real-time
|
|
119
|
+
- Download trained embeddings (latent and interpretable)
|
|
120
|
+
|
|
121
|
+
---
|
|
85
122
|
|
|
86
123
|
### Basic scATAC-seq Workflow
|
|
87
124
|
|
|
@@ -268,7 +305,7 @@ iaode.datasets.clear_cache() # Clear all cached data
|
|
|
268
305
|
|
|
269
306
|
## Examples
|
|
270
307
|
|
|
271
|
-
|
|
308
|
+
Examples are provided in the `examples/` directory. See **[examples/README.md](examples/README.md)** for details.
|
|
272
309
|
|
|
273
310
|
### Example Scripts
|
|
274
311
|
|
|
@@ -555,12 +592,16 @@ for epoch in range(100):
|
|
|
555
592
|
If you use iAODE in your research, please cite:
|
|
556
593
|
|
|
557
594
|
```bibtex
|
|
558
|
-
@
|
|
559
|
-
|
|
560
|
-
|
|
561
|
-
|
|
562
|
-
|
|
563
|
-
|
|
595
|
+
@article{Fu2026iAODE,
|
|
596
|
+
author = {Fu, Zeyu and Chen, Chunlin and Wang, Song and Wang, Junping and Chen, Shilei},
|
|
597
|
+
title = {iAODE for benchmarking and continuum modeling of single-cell chromatin accessibility},
|
|
598
|
+
journal = {Communications Biology},
|
|
599
|
+
volume = {9},
|
|
600
|
+
number = {1},
|
|
601
|
+
pages = {507},
|
|
602
|
+
year = {2026},
|
|
603
|
+
doi = {10.1038/s42003-026-09768-8},
|
|
604
|
+
url = {https://doi.org/10.1038/s42003-026-09768-8}
|
|
564
605
|
}
|
|
565
606
|
```
|
|
566
607
|
|
|
@@ -662,7 +662,9 @@ class agent(Env):
|
|
|
662
662
|
T = adata.obsp[T_key].copy()
|
|
663
663
|
|
|
664
664
|
if self_transition:
|
|
665
|
-
|
|
665
|
+
row_max = T.max(1)
|
|
666
|
+
# Sparse results have no .A since SciPy 1.14; densify explicitly.
|
|
667
|
+
max_t = np.asarray(row_max.todense() if issparse(row_max) else row_max).ravel()
|
|
666
668
|
ub = np.percentile(max_t, 98)
|
|
667
669
|
self_t = np.clip(ub - max_t, 0, 1)
|
|
668
670
|
if hasattr(T, 'setdiag'):
|
|
@@ -877,6 +879,6 @@ def l2_norm(x: np.ndarray, axis: int = -1) -> np.ndarray:
|
|
|
877
879
|
L2 norms
|
|
878
880
|
"""
|
|
879
881
|
if issparse(x):
|
|
880
|
-
return np.sqrt(x.multiply(x).sum(axis=axis).
|
|
882
|
+
return np.sqrt(np.asarray(x.multiply(x).sum(axis=axis)).ravel())
|
|
881
883
|
else:
|
|
882
884
|
return np.sqrt(np.sum(x * x, axis=axis))
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.4
|
|
2
2
|
Name: iaode
|
|
3
|
-
Version: 0.
|
|
3
|
+
Version: 0.3.1
|
|
4
4
|
Summary: Interpretable Autoencoder with Ordinary Differential Equations for single-cell omics analysis
|
|
5
5
|
Author-email: Zeyu Fu <fuzeyu99@126.com>
|
|
6
6
|
License-Expression: MIT
|
|
@@ -31,6 +31,7 @@ Requires-Dist: torchdiffeq>=0.2.0
|
|
|
31
31
|
Requires-Dist: anndata>=0.8.0
|
|
32
32
|
Requires-Dist: scanpy>=1.8.0
|
|
33
33
|
Requires-Dist: scvi-tools>=0.16.0
|
|
34
|
+
Requires-Dist: zarr>=2.0.0
|
|
34
35
|
Requires-Dist: requests>=2.25.0
|
|
35
36
|
Requires-Dist: tqdm>=4.60.0
|
|
36
37
|
Provides-Extra: dev
|
|
@@ -47,18 +48,36 @@ Requires-Dist: sphinx-rtd-theme>=1.0.0; extra == "docs"
|
|
|
47
48
|
Requires-Dist: sphinx-autodoc-typehints>=1.18.0; extra == "docs"
|
|
48
49
|
Dynamic: license-file
|
|
49
50
|
|
|
51
|
+
<div align="center">
|
|
52
|
+
<a href="https://peterponyu.github.io/">
|
|
53
|
+
<img src="https://peterponyu.github.io/assets/badges/iAODE.svg" width="64" alt="ZF Lab · iAODE">
|
|
54
|
+
</a>
|
|
55
|
+
</div>
|
|
56
|
+
|
|
50
57
|
# iAODE: Interpretable Accessibility ODE VAE for scATAC-seq
|
|
51
58
|
|
|
52
|
-
[](https://pypi.org/project/iaode/) [](https://www.python.org/downloads/) [](https://opensource.org/licenses/MIT) [](https://pytorch.org/) [](https://github.com/PeterPonyu/iAODE/actions/workflows/test.yml)
|
|
59
|
+
[](https://pypi.org/project/iaode/) [](https://www.python.org/downloads/) [](https://opensource.org/licenses/MIT) [](https://pytorch.org/) [](https://github.com/PeterPonyu/iAODE/actions/workflows/test.yml) [](https://doi.org/10.5281/zenodo.18453104)
|
|
60
|
+
|
|
61
|
+
> **Interactive Tools:** The dataset browser and continuity explorer have moved to [**SCPortal**](https://peterponyu.github.io/scportal/) — a unified single-cell data portal with expanded datasets, LAIOR benchmarks, and more. The [original project page](https://peterponyu.github.io/iAODE/) is preserved as the manuscript publication snapshot.
|
|
62
|
+
|
|
63
|
+
## Public Pages and local application surfaces
|
|
64
|
+
|
|
65
|
+
`site/` is the public GitHub Pages landing at the canonical URL
|
|
66
|
+
`https://peterponyu.github.io/iAODE/`. It provides project documentation and
|
|
67
|
+
reproducible entry points only. No `/iAODE/frontend/` Pages route is supported.
|
|
68
|
+
|
|
69
|
+
`frontend/` and `api/static/` are local application assets for the training UI.
|
|
70
|
+
They remain local so model training and data handling are performed in the
|
|
71
|
+
user's environment rather than on the public Pages site.
|
|
53
72
|
|
|
54
|
-
**iAODE** (Interpretable Accessibility ODE VAE) is a lightweight deep learning framework
|
|
73
|
+
**iAODE** (Interpretable Accessibility ODE VAE) is a lightweight deep learning framework centered on single-cell ATAC-seq (scATAC-seq) data. It integrates a Variational Autoencoder (VAE) with a Neural ODE and an interpretable bottleneck to support:
|
|
55
74
|
|
|
56
|
-
1. **
|
|
75
|
+
1. **Count-based Modeling:** Handles sparse accessibility profiles with Negative Binomial (NB) or Zero-Inflated Negative Binomial (ZINB) likelihoods.
|
|
57
76
|
2. **Continuous Trajectory Inference:** Infers dynamics via Neural ODE pseudotime and latent velocity fields.
|
|
58
77
|
3. **Interpretable Latent Factors:** Utilizes a biologically aligned bottleneck to extract meaningful features.
|
|
59
78
|
4. **Scalable Preprocessing:** Implements TF‑IDF normalization and Highly Variable Peak (HVP) selection, aligned with Signac and SnapATAC2 best practices.
|
|
60
79
|
|
|
61
|
-
The
|
|
80
|
+
The core package targets scATAC-seq data, including sparse peak counts, library size variability, heterogeneous accessibility kinetics, and dynamic regulatory trajectories. Some bundled scripts and UI workflows also cover scRNA-seq comparison tasks, but scATAC-seq remains the primary scope of the repository.
|
|
62
81
|
|
|
63
82
|
---
|
|
64
83
|
|
|
@@ -98,39 +117,58 @@ Loss = recon + i_recon + ODE_consistency + β·KL + dip·DIP + tc·TC + info·MM
|
|
|
98
117
|
|
|
99
118
|
## Installation
|
|
100
119
|
|
|
101
|
-
###
|
|
102
|
-
|
|
120
|
+
### From PyPI (Recommended)
|
|
121
|
+
|
|
103
122
|
```bash
|
|
104
123
|
pip install iaode
|
|
105
124
|
```
|
|
106
125
|
|
|
107
|
-
###
|
|
108
|
-
For users who want the **Interactive Training UI**, you must clone the repository (since the UI requires the server files):
|
|
126
|
+
### From Source
|
|
109
127
|
|
|
110
128
|
```bash
|
|
111
129
|
git clone https://github.com/PeterPonyu/iAODE.git
|
|
112
130
|
cd iAODE
|
|
113
131
|
pip install -e .
|
|
114
|
-
pip install fastapi uvicorn # Required for the UI server
|
|
115
132
|
```
|
|
116
133
|
|
|
134
|
+
### Requirements
|
|
135
|
+
|
|
136
|
+
- Python ≥ 3.9
|
|
137
|
+
- PyTorch ≥ 1.10.0
|
|
138
|
+
- AnnData ≥ 0.8.0
|
|
139
|
+
- Scanpy ≥ 1.8.0
|
|
140
|
+
- scvi-tools ≥ 0.16.0
|
|
141
|
+
|
|
142
|
+
See `requirements.txt` for the complete list of dependencies.
|
|
143
|
+
|
|
117
144
|
---
|
|
145
|
+
|
|
118
146
|
## Quick Start
|
|
119
147
|
|
|
120
|
-
###
|
|
121
|
-
**No Node.js required.** The interface is pre-built and served via Python.
|
|
148
|
+
### Training UI (Web Interface)
|
|
122
149
|
|
|
123
|
-
|
|
124
|
-
|
|
125
|
-
|
|
126
|
-
|
|
127
|
-
|
|
128
|
-
|
|
129
|
-
|
|
130
|
-
|
|
131
|
-
|
|
132
|
-
|
|
133
|
-
|
|
150
|
+
iAODE includes an integrated web-based training interface for interactive model training:
|
|
151
|
+
|
|
152
|
+
```bash
|
|
153
|
+
# Start the training UI (Python - cross-platform)
|
|
154
|
+
python start_training_ui.py
|
|
155
|
+
|
|
156
|
+
# Or use the shell script (Linux/macOS)
|
|
157
|
+
./start_training_ui.sh
|
|
158
|
+
```
|
|
159
|
+
|
|
160
|
+
Then open your browser to:
|
|
161
|
+
- **Training UI**: http://localhost:8000/ui
|
|
162
|
+
- **API Documentation**: http://localhost:8000/docs
|
|
163
|
+
|
|
164
|
+
**Features:**
|
|
165
|
+
- Upload scATAC-seq data and related `.h5ad` workflows through the UI
|
|
166
|
+
- Configure preprocessing parameters (TF-IDF, HVP selection, subsampling)
|
|
167
|
+
- Set training parameters interactively
|
|
168
|
+
- Monitor training progress in real-time
|
|
169
|
+
- Download trained embeddings (latent and interpretable)
|
|
170
|
+
|
|
171
|
+
---
|
|
134
172
|
|
|
135
173
|
### Basic scATAC-seq Workflow
|
|
136
174
|
|
|
@@ -317,7 +355,7 @@ iaode.datasets.clear_cache() # Clear all cached data
|
|
|
317
355
|
|
|
318
356
|
## Examples
|
|
319
357
|
|
|
320
|
-
|
|
358
|
+
Examples are provided in the `examples/` directory. See **[examples/README.md](examples/README.md)** for details.
|
|
321
359
|
|
|
322
360
|
### Example Scripts
|
|
323
361
|
|
|
@@ -604,12 +642,16 @@ for epoch in range(100):
|
|
|
604
642
|
If you use iAODE in your research, please cite:
|
|
605
643
|
|
|
606
644
|
```bibtex
|
|
607
|
-
@
|
|
608
|
-
|
|
609
|
-
|
|
610
|
-
|
|
611
|
-
|
|
612
|
-
|
|
645
|
+
@article{Fu2026iAODE,
|
|
646
|
+
author = {Fu, Zeyu and Chen, Chunlin and Wang, Song and Wang, Junping and Chen, Shilei},
|
|
647
|
+
title = {iAODE for benchmarking and continuum modeling of single-cell chromatin accessibility},
|
|
648
|
+
journal = {Communications Biology},
|
|
649
|
+
volume = {9},
|
|
650
|
+
number = {1},
|
|
651
|
+
pages = {507},
|
|
652
|
+
year = {2026},
|
|
653
|
+
doi = {10.1038/s42003-026-09768-8},
|
|
654
|
+
url = {https://doi.org/10.1038/s42003-026-09768-8}
|
|
613
655
|
}
|
|
614
656
|
```
|
|
615
657
|
|
|
@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
|
|
|
4
4
|
|
|
5
5
|
[project]
|
|
6
6
|
name = "iaode"
|
|
7
|
-
version = "0.
|
|
7
|
+
version = "0.3.1"
|
|
8
8
|
description = "Interpretable Autoencoder with Ordinary Differential Equations for single-cell omics analysis"
|
|
9
9
|
readme = "README.md"
|
|
10
10
|
authors = [
|
|
@@ -44,6 +44,7 @@ dependencies = [
|
|
|
44
44
|
"anndata>=0.8.0",
|
|
45
45
|
"scanpy>=1.8.0",
|
|
46
46
|
"scvi-tools>=0.16.0",
|
|
47
|
+
"zarr>=2.0.0",
|
|
47
48
|
"requests>=2.25.0",
|
|
48
49
|
"tqdm>=4.60.0",
|
|
49
50
|
]
|
|
@@ -0,0 +1,11 @@
|
|
|
1
|
+
from pathlib import Path
|
|
2
|
+
|
|
3
|
+
|
|
4
|
+
def test_public_page_contains_scpportal_tool_routes() -> None:
|
|
5
|
+
html = (Path(__file__).parents[1] / "site" / "index.html").read_text(encoding="utf-8")
|
|
6
|
+
assert "Dataset Browser" in html
|
|
7
|
+
assert "https://peterponyu.github.io/scportal/datasets/" in html
|
|
8
|
+
assert "Continuity Explorer" in html
|
|
9
|
+
assert "https://peterponyu.github.io/scportal/explorer/" in html
|
|
10
|
+
assert "localhost" not in html.lower()
|
|
11
|
+
assert "/iAODE/frontend/" not in html
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|