Deep Learning-based Clustering Approaches for Bioinformatics

Codes and supplementary materials for our paper "Deep Learning-based Clustering Approaches for Bioinformatics" published in Briefings in Bioinformatics journal. This repo will be updated periodically. In particular, more complete Jupyter notebooks will be added. In this article, we reviewed deep learning-based approaches for cluster analysis, including network training, representation learning, parameter optimization, and formulating clustering quality metrics. We also discussed how representation learning based on different autoencoder architectures (e.g., vanilla, variational, LSTM, and convolutional) can be more effective than ML-based approaches (e.g., PCA) in different scenarios, e.g., bio-imaging, gene expression clustering, and clustering biomedical texts.

Deep learning-based unsupervised/clustering methods, link to papers & codes

We provide the list of deep learning-based unsupervised/clustering methods, link to papers, and codes. Besides, new articles proposing approaches and paper will be listed. So stay tuned!

Title	Article	Conference/Journal	Code
Deep clustering with convolutional autoencoders (DCEC)	Link	ICONIP'2017	GitHub
Unsupervised Data Augmentation for Consistency Training (UDA)	Link	Arxiv'2019	GitHub
Deep Clustering via joint convolutional autoencoder embedding and relative entropy minimization (DEPICT)	Link	ICCV'2017	GitHub
Discriminatively Boosted Clustering (DBC)	Link	Arxiv'2017	N/A
Variational Deep Embedding (VADE)	Link	IJCAI'2017	GitHub
Convolutional Embedded Networks (CEN)}	Link	Arxiv'2018	GitHub
Deep Subspace Clustering Networks (DSC-Nets)	Link	NIPS'2017	GitHub
Graph Clustering with Dynamic Embedding (GRACE)	Link	Arxiv'2017	N/A
Deep Unsupervised Clustering Using Mixture of Autoencoders (MIXAE)	Link	Arxiv'2017	N/A
Deep Embedded Clustering (DEC)	Link	ICML'2016	GitHub
A Survey of Clustering With Deep Learning: From the Perspective of Network Architecture	Link	IEEE ACCESS 2018
GEMSEC: Graph Embedding with Self Clustering	Link	Arxiv,2018	GitHub
Clustering with Deep Learning: Taxonomy and New Methods	Link	Arxiv, 2018	GitHub
Deep Continuous Clustering (DCC)	Link	Arxiv, 2018	GitHub
Deep Clustering with Convolutional Autoencoders (DCEC)	Link	ICONIP'2018	GitHub
SpectralNet: Spectral Clustering Using Deep Neural Networks	Link	ICLR'2018	GitHub
Subspace clustering using a low-rank constrained autoencoder (LRAE)	Link	Information Sciences'2018	N/A
Clustering-driven Deep Embedding with Pairwise Constraints (CPAC)	Link	Arxiv'2018	GitHub
Towards K-means-friendly Spaces: Simultaneous Deep Learning and Clustering	Link	PMLR'2017	N/A
Deep Unsupervised Clustering With Gaussian Mixture Variational AutoEncoders (GMVAE)	Link	ICLR'2017	GitHub
Is Simple Better?: Revisiting Simple Generative Models for Unsupervised Clustering	Link	NIPS'2017 Workshop	GitHub
Imporved Deep Embedding Clustering (IDEC)	Link	IJCAI'2017	GitHub
Deep Clustering Network (DCN)	Link	Arxiv'2016	GitHub
Joint Unsupervised Learning of Deep Representations and Image Clustering (JULE)	Link	CVPR'2016	GitHub
Deep Embedding Network for Clustering (DEN)	Link	ICPR'2014	N/A
Auto-encoder Based Data Clustering (ABDC)	Link	CIARP'2013	GitHub
Learning Deep Representations for Graph Clustering	Link	AAAI'2014	GitHub

Running provided Jupyter notebooks

To run the examples interactively, you need to install some Python modules and libraries.

• Python 3
• Scikit-learn
• Keras
• TensorFlow.

For the Jupyter notebook, git it from this Link and install it on your machine. Then clone this repo using following command, given that you have already installed the git:

git clone https://github.com/rezacsedu/Deep-learning-for-clustering-in-bioinformatics.git

Alternatively, install all the required libraries by issuing the following command:

 cd Deep-learning-for-clustering-in-bioinformatics
 pip3 install -r requirements.txt
 cd Notebboks

Then start Jupyter notebbok by issuing the following command:

jupyter notebook

In the opened browser, go to Jupyter tab and window open the notebook.

LSTM_AE_Text_Clustering.ipynb

If you want to skip the training, soon we'll provide the pre-trained weights, which you can restore and start fine-tuning. Happy coding! Leave a comment if you have any question.

Acknowledgement

The ClusteringLayer class and the target_distribution function are based on DEC from https://github.com/XifengGuo/DCEC/blob/master/DCEC.py by Xifeng Guo

Citation request

If you use the code of this repository in your research, please consider citing the folowing papers:

@article{karim2021deep,
      title={Deep learning-based clustering approaches for bioinformatics},
      author={Karim, Md Rezaul and Beyan, Oya and Zappa, Achille and Costa, Ivan G and Rebholz-Schuhmann, Dietrich and Cochez, Michael and Decker, Stefan},
      journal={Briefings in bioinformatics},
      volume={22},
      number={1},
      pages={393--415},
      year={2021},
      publisher={Oxford University Press}
      }

Contributing

If you find more related work, which are not listed here, please create a PR or sugest by filing issues. Your contribution will be highly appreciated. For any questions, feel free to open an issue or contact at [email protected].