Sparsely Connected Autoencoders: A Multi-Purpose Tool for Single Cell omics Analysis

Abstract

Background: Biological processes are based on complex networks of cells and molecules. Single cell multi-omics is a new tool aiming to provide new incites in the complex network of events controlling the functionality of the cell. Methods: Since single cell technologies provide many sample measurements, they are the ideal environment for the application of Deep Learning and Machine Learning approaches. An autoencoder is composed of an encoder and a decoder sub-model. An autoencoder is a very powerful tool in data compression and noise removal. However, the decoder model remains a black box from which is impossible to depict the contribution of the single input elements. We have recently developed a new class of autoencoders, called Sparsely Connected Autoencoders (SCA), which have the advantage of providing a controlled association among the input layer and the decoder module. This new architecture has the benefit that the decoder model is not a black box anymore and can be used to depict new biologically interesting features from single cell data. Results: Here, we show that SCA hidden layer can grab new information usually hidden in single cell data, like providing clustering on meta-features difficult, i.e. transcription factors expression, or not technically not possible, i.e. miRNA expression, to depict in single cell RNAseq data. Furthermore, SCA representation of cell clusters has the advantage of simulating a conventional bulk RNAseq, which is a data transformation allowing the identification of similarity among independent experiments. Conclusions: In our opinion, SCA represents the bioinformatics version of a universal “Swiss-knife” for the extraction of hidden knowledgeable features from single cell omics data.

ZINBMM: a general mixture model for simultaneous clustering and gene selection using single-cell transcriptomic data

“…Gene Trh is a unique endoderm marker that transiently marks the entire definitive endoderm population and is not expressed in the extraembryonic endoderm. In the human lung cell dataset, gene CD74 has been shown to play an important role in eliciting immune response in lung adenocarcinoma [ 42 ]. CDKN2A has been identified as a tumour suppressor associated with the detection of regulatory gene hubs [ 42 ].…”

“…In the human lung cell dataset, gene CD74 has been shown to play an important role in eliciting immune response in lung adenocarcinoma [ 42 ]. CDKN2A has been identified as a tumour suppressor associated with the detection of regulatory gene hubs [ 42 ]. Gene EGFR has been commonly used as an important therapeutic target for non-small-cell lung carcinoma (NSCLC) [ 43 ].…”

NetActivity enhances transcriptional signals by combining gene expression into robust gene set activity scores through interpretable autoencoders

“…Shallow sparsely-connected autoencoders pose as ideal alternatives to address the shortcomings of current gene set projection methods [14][15][16] . Autoencoders are neural networks (NN) where the input data is reduced to a lower dimension and then expanded to reconstruct the original input data.…”

“…Recently, shallow sparsely-connected autoencoders have been proposed to define GSAS [14][15][16] , where each gene set is represented by a neuron of the inner layer and is only connected to the genes in the gene set. The model aims to learn the low-dimensional embedding (i.e., the set of gene set scores) that best represents the input data.…”

Editorial of Special Issue “Deep Learning and Machine Learning in Bioinformatics”

“…Nosi et al proposed a neural network method to detect MET exon 14 skipping events using RNAseq data from The Cancer Genome Atlas (TCGA) archive for lung cancer [ 4 ]. Alessandri et al developed a new autoencoder model, called Sparsely Connected Autoencoders, to improve the traditional decoder model for better identifying biological features from single cell data [ 5 ]. Al Mamun et al developed a multi-run concrete autoencoder to identify a stable set of features which was applied to TCGA genome-wide lncRNA expression profiles in 12 cancers, resulting in the identification of key lncRNAs [ 6 ].…”