SMILE: mutual information learning for integration of single-cell omics data

Xu, Yang; Das, Priyojit; McCord, Rachel Patton

doi:10.1093/bioinformatics/btab706

Cited by 37 publications

(33 citation statements)

References 56 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Instead of calculating NCE directly on z , we further reduced z to 32-dimension output with linear transformation and 25-dimension SoftMax activated output, through two separated one-layer MLPs. This practice is the same as our previous study, in which we demonstrated an effective approach to learn discriminative representation for single-cell data 22 . Once model training is done, we use encoder E to project both modalities into the joint representation for downstream analyses (Fig.…”

Section: Resultsmentioning

confidence: 91%

“…We term our method sciCAN (single-cell chromatin accessibility and gene expression data i ntegration via Cycle-consistent Adversarial Network), which removes modality differences while keeping true biological variation. We previously developed a deep learning method, SMILE, to perform integration of multimodal single-cell data 22 . SMILE requires cell anchors for integration.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

sciCAN: single-cell chromatin accessibility and gene expression data integration via cycle-consistent adversarial network

Begoli

McCord

2022

npj Syst Biol Appl

Self Cite

View full text Add to dashboard Cite

The boom in single-cell technologies has brought a surge of high dimensional data that come from different sources and represent cellular systems from different views. With advances in these single-cell technologies, integrating single-cell data across modalities arises as a new computational challenge. Here, we present an adversarial approach, sciCAN, to integrate single-cell chromatin accessibility and gene expression data in an unsupervised manner. We benchmarked sciCAN with 5 existing methods in 5 scATAC-seq/scRNA-seq datasets, and we demonstrated that our method dealt with data integration with consistent performance across datasets and better balance of mutual transferring between modalities than the other 5 existing methods. We further applied sciCAN to 10X Multiome data and confirmed that the integrated representation preserves biological relationships within the hematopoietic hierarchy. Finally, we investigated CRISPR-perturbed single-cell K562 ATAC-seq and RNA-seq data to identify cells with related responses to different perturbations in these different modalities.

show abstract

Section: Resultsmentioning

confidence: 91%

Section: Introductionmentioning

confidence: 99%

sciCAN: single-cell chromatin accessibility and gene expression data integration via cycle-consistent adversarial network

Begoli

McCord

2022

npj Syst Biol Appl

Self Cite

View full text Add to dashboard Cite

show abstract

“…#9 pipeline is a combination process of #2 and #8 pipelines. Deep-learning-based batch correction methods demonstrated a considerable success to integrative analysis of scRNA-seq data, and we noticed that the frequent practice across these methods is use of batch normalization layer and non-linear activation layer, which splits the whole dataset into multiple mini-batches, standardizes cells in each batch, and transforms the outcome with a non-linear activation function 2,4,5,13,14 . This batch normalization and non-linear activation process do not require weight training, and we included it into the #10 pipeline.…”

Section: Resultsmentioning

confidence: 99%

Fast model-free standardization and integration of single-cell transcriptomics data

Kramann

McCord

Hayat

2022

Preprint

Self Cite

View full text Add to dashboard Cite

Single-cell transcriptomics datasets from the same anatomical sites generated by different research labs are becoming mainstream. However, fast, and computationally inexpensive tools for standardization of cell-type annotation and data integration are still needed to increase research inclusivity. To standardize cell-type annotation and integrate single-cell transcriptomics datasets, we have built a fast, model-free integration method called MASI (Marker-Assisted Standardization and Integration). MASI can run integrative annotation on a personal laptop for approximately one million cells, providing a cheap computational alternative for the single-cell data analysis community. MASI has an average macro F1/overall accuracy of 0.79/0.89 over the 4 benchmark datasets. We demonstrate that MASI outperforms other methods based on speed, and its performance for the tasks of data integration and cell-type annotation is comparable or even superior to other existing methods. We apply MASI for integrative lineage analysis and show that it preserves the underlying biological signal in datasets tested. Finally, to harness knowledge from single-cell atlases, we demonstrate three case studies that cover integration across research groups, biological conditions, and surveyed participants, respectively.

show abstract

“…Single cell technology is becoming every day more used to investigate complex questions. Thus, it is becoming important to integrate the results of multiple experiments [ 18 , 20 , 21 ]. Seurat-based integration [ 18 ] is an approach frequently used to aggregate different single cells or different modalities of the same experiment.…”

Section: Resultsmentioning

confidence: 99%

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

Alessandrì

Ratto

Contaldo

et al. 2021

IJMS

View full text Add to dashboard Cite

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.

show abstract

SMILE: mutual information learning for integration of single-cell omics data

Cited by 37 publications

References 56 publications

sciCAN: single-cell chromatin accessibility and gene expression data integration via cycle-consistent adversarial network

sciCAN: single-cell chromatin accessibility and gene expression data integration via cycle-consistent adversarial network

Fast model-free standardization and integration of single-cell transcriptomics data

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

Contact Info

Product

Resources

About