Data

Tools

Indexes

Applications

Experiments

Open Source Science

aggmap

Jigsaw-like AggMap: A Robust and Explainable Multi-Channel Omics Deep Learning Tool

https://github.com/shenwanxiang/bidd-aggmap

Science Score: 36.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ○
    CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ✓
    .zenodo.json file
    Found .zenodo.json file
  • ○
    DOI references
  • ○
    Academic publication links
  • ✓
    Committers with academic emails
    1 of 7 committers (14.3%) from academic institutions
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (11.2%) to scientific vocabulary

Keywords

ai clustering deep-learning machine-learning omics visualization

Keywords from Contributors

mesh sequences interactive hacking network-simulation
Last synced: 6 months ago · JSON representation

Repository

Jigsaw-like AggMap: A Robust and Explainable Multi-Channel Omics Deep Learning Tool

Basic Info
Statistics
  • Stars: 34
  • Watchers: 1
  • Forks: 5
  • Open Issues: 2
  • Releases: 4
Topics
ai clustering deep-learning machine-learning omics visualization
Created over 5 years ago · Last pushed almost 2 years ago
Metadata Files
Readme License Citation Security

README.md

Example Documentation Status Downloads PyPI version Paper

Jigsaw-like AggMap

A Robust and Explainable Omics Deep Learning Tool

Installation (Only on Linux system)

install aggmap by: ```bash

create an aggmap env

conda create -n aggmap python=3.8 conda activate aggmap pip install --upgrade pip pip install aggmap==1.2.1 ```

Usage

```python import pandas as pd from sklearn.datasets import loadbreastcancer from aggmap import AggMap, AggMapNet

Data loading

data = loadbreastcancer() dfx = pd.DataFrame(data.data, columns=data.featurenames) dfy = pd.getdummies(pd.Series(data.target))

AggMap object definition, fitting, and saving

mp = AggMap(dfx, metric = 'correlation') mp.fit(clusterchannels=5, embmethod = 'umap', verbose=0) mp.save('agg.mp')

AggMap visulizations: Hierarchical tree, embeddng scatter and grid

mp.plottree() mp.plotscatter(enableddatalabels=True, radius=5) mp.plotgrid(enableddata_labels=True)

Transoformation of 1d vectors to 3D Fmaps (-1, w, h, c) by AggMap

X = mp.batchtransform(dfx.values, njobs=4, scale_method = 'minmax') y = dfy.values

AggMapNet training, validation, early stopping, and saving

clf = AggMapNet.MultiClassEstimator(epochs=50, gpuid=0) clf.fit(X, y, Xvalid=None, yvalid=None) clf.save_model('agg.model')

Model explaination by simply-explainer: global, local

simpexplainer = AggMapNet.simplyexplainer(clf, mp) globalsimpimportance = simpexplainer.globalexplain(clf.X, clf.y) localsimpimportance = simpexplainer.localexplain(clf.X[[0]], clf.y[[0]])

Model explaination by shapley-explainer: global, local

shapexplainer = AggMapNet.shapleyexplainer(clf, mp) globalshapimportance = shapexplainer.globalexplain(clf.X) localshapimportance = shapexplainer.localexplain(clf.X[[0]]) ```

How It Works?

  • AggMap flowchart of feature mapping and agglomeration into ordered (spatially correlated) multi-channel feature maps (Fmaps)

how-it-works a, AggMap flowchart of feature mapping and aggregation into ordered (spatially-correlated) channel-split feature maps (Fmaps).b, CNN-based AggMapNet architecture for Fmaps learning. c, proof-of-concept illustration of AggMap restructuring of unordered data (randomized MNIST) into clustered channel-split Fmaps (reconstructed MNIST) for CNN-based learning and important feature analysis. d, typical biomedical applications of AggMap in restructuring omics data into channel-split Fmaps for multi-channel CNN-based diagnosis and biomarker discovery (explanation saliency-map of important features).

Proof-of-Concepts of reconstruction ability on MNIST Dataset

  • It can reconstruct to the original image from completely randomly permuted (disrupted) MNIST data:

reconstruction

Org1: the original grayscale images (channel = 1), OrgRP1: the randomized images of Org1 (channel = 1), RPAgg1, 5: the reconstructed images of OrgPR1 by AggMap feature restructuring (channel = 1, 5 respectively, each color represents features of one channel). RPAgg5-tkb: the original images with the pixels divided into 5 groups according to the 5-channels of RPAgg5 and colored in the same way as RPAgg5.

The effect of the number of channels on model performance

  • Multi-channel Fmaps can boost the model performance notably: channel_effect

The performance of AggMapNet using different number of channels on the TCGA-T (a) and COV-D (b). For TCGA-T, ten-fold cross validation average performance, for COV-D, a fivefold cross validation was performed and repeat 5 rounds using different random seeds (total 25 training times), their average performances of the validation set were reported.

Example for Restructured Fmaps

  • The example on WDBC dataset: click here to find out more! Fmap

Citation

Shen, Wan Xiang, et al. "AggMapNet: enhanced and explainable low-sample omics deep learning with feature-aggregated multi-channel networks." Nucleic Acids Research 50.8 (2022): e45-e45.

Owner

  • Name: Charleshen
  • Login: shenwanxiang
  • Kind: user
  • Location: Singapore
  • Company: FoS, National University of Singapore

A data scientist, interested in Bioinformatics & Chemoinformatics

GitHub Events

Total
  • Watch event: 4
  • Fork event: 2
Last Year
  • Watch event: 4
  • Fork event: 2

Committers

Last synced: over 1 year ago

All Time
  • Total Commits: 377
  • Total Committers: 7
  • Avg Commits per committer: 53.857
  • Development Distribution Score (DDS): 0.427
Past Year
  • Commits: 21
  • Committers: 2
  • Avg Commits per committer: 10.5
  • Development Distribution Score (DDS): 0.048
Top Committers
Name Email Commits
shenwanxiang s****g@t****n 216
Charleshen s****g@1****m 86
shenwanxiang s****g@t****n 61
shenwanxiang s****g@g****m 7
shenwanxiang h****g@g****m 5
Shen w****6@c****u 1
dependabot[bot] 4****] 1
Committer Domains (Top 20 + Academic)
github.com: 2 tsinghua.org.cn: 2 compute-e-16-237.o2.rc.hms.harvard.edu: 1 163.com: 1

Issues and Pull Requests

Last synced: over 1 year ago

All Time
  • Total issues: 6
  • Total pull requests: 12
  • Average time to close issues: 13 days
  • Average time to close pull requests: about 2 months
  • Total issue authors: 5
  • Total pull request authors: 2
  • Average comments per issue: 2.17
  • Average comments per pull request: 0.67
  • Merged pull requests: 3
  • Bot issues: 0
  • Bot pull requests: 9
Past Year
  • Issues: 0
  • Pull requests: 0
  • Average time to close issues: N/A
  • Average time to close pull requests: N/A
  • Issue authors: 0
  • Pull request authors: 0
  • Average comments per issue: 0
  • Average comments per pull request: 0
  • Merged pull requests: 0
  • Bot issues: 0
  • Bot pull requests: 0
View more stats
Top Authors
Issue Authors
  • shenwanxiang (2)
  • computbiol (1)
  • ivy-yuan (1)
  • tttender (1)
  • dxbdxx (1)
Pull Request Authors
  • dependabot[bot] (9)
  • shenwanxiang (3)
Top Labels
Issue Labels
dependencies (3) good first issue (1) enhancement (1)
Pull Request Labels
dependencies (9)

Packages

  • Total packages: 1
  • Total downloads:
    • pypi 121 last-month
  • Total dependent packages: 0
  • Total dependent repositories: 1
  • Total versions: 10
  • Total maintainers: 1
pypi.org: aggmap

Jigsaw-like AggMap: A Robust and Explainable Omics Deep Learning Tool

  • Versions: 10
  • Dependent Packages: 0
  • Dependent Repositories: 1
  • Downloads: 121 Last month
Rankings
Dependent packages count: 10.1%
Stargazers count: 11.2%
Forks count: 13.3%
Average: 15.5%
Downloads: 21.3%
Dependent repos count: 21.6%
Maintainers (1)
shenwanxiang
Last synced: 6 months ago

Dependencies

docs/doc_requirements.txt pypi
  • IPython *
  • Sphinx ==5.1.1
  • bokeh *
  • ipykernel *
  • nbsphinx *
  • sphinx-rtd-theme *
  • sphinx_gallery *