Robotic data acquisition with deep learning enables cell image–based prediction of transcriptomic phenotypes

Abstract

Single-cell whole-transcriptome analysis is the gold standard approach to identifying molecularly defined cell phenotypes. However, this approach cannot be used for dynamics measurements such as live-cell imaging. Here, we developed a multifunctional robot, the automated live imaging and cell picking system (ALPS) and used it to perform single-cell RNA sequencing for microscopically observed cells with multiple imaging modes. Using robotically obtained data that linked cell images and the whole transcriptome, we successfully predicted transcriptome-defined cell phenotypes in a noninvasive manner using cell image–based deep learning. This noninvasive approach opens a window to determine the live-cell whole transcriptome in real time. Moreover, this work, which is based on a data-driven approach, is a proof of concept for determining the transcriptome-defined phenotypes (i.e., not relying on specific genes) of any cell from cell images using a model trained on linked datasets.

Opto-combinatorial indexing enables high-content transcriptomics by linking cell images and transcriptomes

“…12 As an example of the latter strategy, an automated cell-picking system was employed to isolate single cells into 96-well plates and then process them for scRNA-seq. 20…”

“…12 As an example of the latter strategy, an automated cell-picking system was employed to isolate single cells into 96-well plates and then process them for scRNA-seq. 20 In contrast to these methods, our approach to link the cellular phenotype with transcriptomics uses a combination of colour codes of cells and hydrogel beads to optically index pairs of single cells and hydrogel beads. Automated microfluidic controls and robotic systems are not required for this method, which incorporates fewer on-chip steps, and is compatible with standard epi-fluorescence microscopy, which are advantageous features because they can be implemented in a standard laboratory.…”

Pigmentation level of human iPSC-derived RPE does not indicate a specific gene expression profile

“…Live cell imaging, cell picking, and single-cell digital RNA-seq ( Shiroguchi et al, 2012 ; Ogawa et al, 2017 ) were performed as described previously ( Jin et al, 2023 ) except for the following: In total, 4032 cells ( Figure 1—source data 1 ) were measured and analyzed. Cell images (400 pixels ×330 pixels; 0.36 μm/pixel) were captured by both bright field and fluorescent channels (filter unit: mCherry C-FLL-C, Nikon Co., Tokyo, Japan) with 20X objective (N.A.…”

Pigmentation level of human iPSC-derived retinal pigment epithelium cell does not indicate a specific gene expression profile

“…Live cell imaging, cell picking, and single cell digital RNA-seq (Shiroguchi et al, 2012; Ogawa et al, 2017) were performed as described previously (Jin et al, 2023) except for the followings: In total, 4032 cells ( Figure 1 –table 1 ) were measured and analyzed. Cell images (400 pixel × 330 pixel; 0.36 μm/pixel) were captured by both bright field and fluorescent channels (filter unit: mCherry C-FLL-C, Nikon Co.) with 20× objective (N.A.…”

“…Live cell imaging, cell picking, and single cell digital RNA-seq (Shiroguchi et al, 2012;Ogawa et al, 2017) were performed as described previously (Jin et al, 2023) except for the followings: In total, 4032 cells (Figure 1 -table 1) were measured and analyzed.…”