Phenotype prediction from single-cell RNA-seq data using attention-based neural networks

Abstract

Motivation: A patient's disease phenotype can be driven and determined by specific groups of cells whose marker genes are either unknown or can only be detected at late-stage using conventional bulk assays such as RNA-Seq technology. Recent advances in single-cell RNA sequencing (scRNA-seq) enable gene expression profiling in cell-level resolution, and therefore have the potential to identify those cells driving the disease phenotype even while the number of these cells is small. However, most existing methods rely heavily on accurate cell type detection, and the number of available annotated samples is usually too small for training deep learning predictive models.

Results: Here, we propose the method ScRAT for phenotype prediction using scRNA-seq data. To train ScRAT with a limited number of samples of different phenotypes, such as coronavirus disease (COVID) and non-COVID, ScRAT first applies a mixup module to increase the number of training samples. A multi-head attention mechanism is employed to learn the most informative cells for each phenotype without relying on a given cell type annotation. Using three public COVID datasets, we show that ScRAT outperforms other phenotype prediction methods. The performance edge of ScRAT over its competitors increases as the number of training samples decreases, indicating the efficacy of our sample mixup. Critical cell types detected based on high-attention cells also support novel findings in the original papers and the recent literature. This suggests that ScRAT overcomes the challenge of missing marker genes and limited sample number with great potential revealing novel molecular mechanisms and/or therapies.

Availability and implementation: The code of our proposed method ScRAT is published at https://github.com/yuzhenmao/ScRAT.

Figure 1.

An overview of ScRAT, which consists of three main modules: Sample Mixup, Attention Layer, and Phenotype Classifier. It takes a scRNA-seq sample (a set of cells) as input, and outputs the predicted phenotype for the input sample.

Figure 2.

Comparison of different methods on four different tasks. For each task, we report the prediction results of all methods using AUC $\pm$ 95% confidence intervals for 10 different training ratios. ScRAT outperforms other methods in all settings, followed by vanilla attention (the P-value of t-test between ScRAT and vanilla attention $\ll$0.01 in all but the SC4-Severity tasks at Training Ratio = 9%). The performance edge of ScRAT over vanilla attention increases as the training ratio decreases, especially for the Combat datasets. See Supplementary Figs S3 and S4 for more information.