DeLTA 2.0: A deep learning pipeline for quantifying single-cell spatial and temporal dynamics

Improvements in microscopy software and hardware have dramatically increased the pace of image acquisition, making analysis a major bottleneck in generating quantitative, single-cell data. Although tools for segmenting and tracking bacteria within time-lapse images exist, most require human input, are specialized to the experimental set up, or lack accuracy. Here, we introduce DeLTA 2.0, a purely Python workflow that can rapidly and accurately analyze single cells on two-dimensional surfaces to quantify gene expression and cell growth. The algorithm uses deep convolutional neural networks to extract single-cell information from time-lapse images, requiring no human input after training. DeLTA 2.0 retains all the functionality of the original version, which was optimized for bacteria growing in the mother machine microfluidic device, but extends results to two-dimensional growth environments. Two-dimensional environments represent an important class of data because they are more straightforward to implement experimentally, they offer the potential for studies using co-cultures of cells, and they can be used to quantify spatial effects and multi-generational phenomena. However, segmentation and tracking are significantly more challenging tasks in two-dimensions due to exponential increases in the number of cells that must be tracked. To showcase this new functionality, we analyze mixed populations of antibiotic resistant and susceptible cells, and also track pole age and growth rate across generations. In addition to the two-dimensional capabilities, we also introduce several major improvements to the code that increase accessibility, including the ability to accept many standard microscopy file formats and arbitrary image sizes as inputs. DeLTA 2.0 is rapid, with run times of less than 10 minutes for complete movies with hundreds of cells, and is highly accurate, with error rates around 1%, making it a powerful tool for analyzing time-lapse microscopy data.

Download Full-text

BCM3D 2.0: Accurate segmentation of single bacterial cells in dense biofilms using computationally generated intermediate image representations

10.1101/2021.11.26.470109 ◽

2021 ◽

Author(s):

Ji Zhang ◽

Yibo Wang ◽

Eric Donarski ◽

Andreas Gahlmann

Keyword(s):

Image Analysis ◽

Bacterial Cell ◽

Single Cells ◽

Three Dimensional ◽

Time Lapse ◽

Cell Segmentation ◽

Bacterial Cells ◽

Deep Convolutional Neural Networks ◽

Fluorescence Images ◽

Image Representations

Accurate detection and segmentation of single cells in three-dimensional (3D) fluorescence time-lapse images is essential for measuring individual cell behaviors in large bacterial communities called biofilms. Recent progress in machine-learning-based image analysis is providing this capability with every increasing accuracy. Leveraging the capabilities of deep convolutional neural networks (CNNs), we recently developed bacterial cell morphometry in 3D (BCM3D), an integrated image analysis pipeline that combines deep learning with conventional image analysis to detect and segment single biofilm-dwelling cells in 3D fluorescence images. While the first release of BCM3D (BCM3D 1.0) achieved state-of-the-art 3D bacterial cell segmentation accuracies, low signal-to-background ratios (SBRs) and images of very dense biofilms remained challenging. Here, we present BCM3D 2.0 to address this challenge. BCM3D 2.0 is completely complementary to the approach utilized in BCM3D 1.0. Instead of training CNNs to perform voxel classification, we trained CNNs to translate 3D fluorescence images into intermediate 3D image representations that are, when combined appropriately later, more amenable to conventional mathematical image processing than a single experimental image. Using this approach, improved segmentation results are obtained even for very low SBRs and/or high cell density biofilm images. The improved cell segmentation accuracies in turn enable improved accuracies of tracking individual cells through 3D space and time, which opens the door to investigating time-dependent phenomena in bacterial biofilms at the cellular level.

Download Full-text

Weakly Supervised Learning of Single-Cell Feature Embeddings

10.1101/293431 ◽

2018 ◽

Cited By ~ 3

Author(s):

Juan C. Caicedo ◽

Claire McQuin ◽

Allen Goodman ◽

Shantanu Singh ◽

Anne E. Carpenter

Keyword(s):

Single Cell ◽

Single Cells ◽

Feature Learning ◽

Experimental Treatment ◽

Ground Truth ◽

Learning Approaches ◽

Deep Convolutional Neural Networks ◽

Experimental Conditions ◽

Weakly Supervised ◽

Experimental Treatments

AbstractWe study the problem of learning representations for single cells in microscopy images to discover biological relationships between their experimental conditions. Many new applications in drug discovery and functional genomics require capturing the morphology of individual cells as comprehensively as possible. Deep convolutional neural networks (CNNs) can learn powerful visual representations, but require ground truth for training; this is rarely available in biomedical profiling experiments. While we do not know which experimental treatments produce cells that look alike, we do know that cells exposed to the same experimental treatment should generally look similar. Thus, we explore training CNNs using a weakly supervised approach that uses this information for feature learning. In addition, the training stage is regularized to control for unwanted variations using mixup or RNNs. We conduct experiments on two different datasets; the proposed approach yields single-cell embeddings that are more accurate than the widely adopted classical features, and are competitive with previously proposed transfer learning approaches.

Download Full-text

Massively parallel and time-resolved RNA sequencing in single cells with scNT-Seq

10.21203/rs.3.pex-1019/v1 ◽

2021 ◽

Author(s):

Qi Qiu ◽

Peng Hu ◽

Hao Wu

Keyword(s):

Single Cell ◽

Rna Sequencing ◽

Temporal Dynamics ◽

Single Cells ◽

Chemical Conversion ◽

Droplet Microfluidics ◽

Massively Parallel ◽

One Pot ◽

Time Resolved ◽

Rna Biogenesis

Abstract Single-cell RNA sequencing offers snapshots of whole transcriptomes but obscures the temporal dynamics of RNA biogenesis and decay. Here we present single-cell metabolically labeled new RNA tagging sequencing (scNT-Seq), a method for massively parallel analysis of newly-transcribed and pre-existing RNAs from the same cell. This droplet microfluidics-based method enables high-throughput chemical conversion on barcoded beads, efficiently marking newly-transcribed RNAs with T-to-C substitutions. The steps of the protocol are (1) metabolically labeling of cells with 4sU, (2) co-encapsulating individual cell with a barcoded oligo-dT primer coated bead in a nanoliter-scale droplet, (3) performing one-pot 4sU chemical conversion on pooled barcoded beads, and (4) reverse transcription, cDNA amplification, tagmentation, indexing PCR, and sequencing. scNT-Seq provides a broadly applicable strategy to investigate dynamic biological systems at single-cell resolution.

Download Full-text

Ultraaccurate genome sequencing and haplotyping of single human cells

Proceedings of the National Academy of Sciences ◽

10.1073/pnas.1707609114 ◽

2017 ◽

Vol 114 (47) ◽

pp. 12512-12517 ◽

Cited By ~ 18

Author(s):

Wai Keung Chu ◽

Peter Edge ◽

Ho Suk Lee ◽

Vikas Bansal ◽

Vineet Bafna ◽

...

Keyword(s):

Single Cell ◽

Genome Sequencing ◽

Single Cells ◽

Error Rates ◽

Human Cells ◽

Sequence Information ◽

Dna Fragments ◽

Chromosomal Dna ◽

Haplotype Information

Accurate detection of variants and long-range haplotypes in genomes of single human cells remains very challenging. Common approaches require extensive in vitro amplification of genomes of individual cells using DNA polymerases and high-throughput short-read DNA sequencing. These approaches have two notable drawbacks. First, polymerase replication errors could generate tens of thousands of false-positive calls per genome. Second, relatively short sequence reads contain little to no haplotype information. Here we report a method, which is dubbed SISSOR (single-stranded sequencing using microfluidic reactors), for accurate single-cell genome sequencing and haplotyping. A microfluidic processor is used to separate the Watson and Crick strands of the double-stranded chromosomal DNA in a single cell and to randomly partition megabase-size DNA strands into multiple nanoliter compartments for amplification and construction of barcoded libraries for sequencing. The separation and partitioning of large single-stranded DNA fragments of the homologous chromosome pairs allows for the independent sequencing of each of the complementary and homologous strands. This enables the assembly of long haplotypes and reduction of sequence errors by using the redundant sequence information and haplotype-based error removal. We demonstrated the ability to sequence single-cell genomes with error rates as low as 10−8 and average 500-kb-long DNA fragments that can be assembled into haplotype contigs with N50 greater than 7 Mb. The performance could be further improved with more uniform amplification and more accurate sequence alignment. The ability to obtain accurate genome sequences and haplotype information from single cells will enable applications of genome sequencing for diverse clinical needs.

Download Full-text

Ultra-accurate Genome Sequencing and Haplotyping of Single Human Cells

10.1101/135384 ◽

2017 ◽

Author(s):

Wai Keung Chu ◽

Peter Edge ◽

Ho Suk Lee ◽

Vikas Bansal ◽

Vineet Bafna ◽

...

Keyword(s):

Single Cell ◽

Genome Sequencing ◽

Single Cells ◽

Error Rates ◽

Human Cells ◽

Sequence Information ◽

Dna Fragments ◽

Chromosomal Dna ◽

Haplotype Information

AbstractAccurate detection of variants and long-range haplotypes in genomes of single human cells remains very challenging. Common approaches require extensive in vitro amplification of genomes of individual cells using DNA polymerases and high-throughput short-read DNA sequencing. These approaches have two notable drawbacks. First, polymerase replication errors could generate tens of thousands of false positive calls per genome. Second, relatively short sequence reads contain little to no haplotype information. Here we report a method, which is dubbed SISSOR (Single-Stranded Sequencing using micrOfluidic Reactors), for accurate single-cell genome sequencing and haplotyping. A microfluidic processor is used to separate the Watson and Crick strands of the double-stranded chromosomal DNA in a single cell and to randomly partition megabase-size DNA strands into multiple nanoliter compartments for amplification and construction of barcoded libraries for sequencing. The separation and partitioning of large single-stranded DNA fragments of the homologous chromosome pairs allows for the independent sequencing of each of the complementary and homologous strands. This enables the assembly of long haplotypes and reduction of sequence errors by using the redundant sequence information and haplotype-based error removal. We demonstrated the ability to sequence single-cell genomes with error rates as low as 10−8 and average 500kb long DNA fragments that can be assembled into haplotype contigs with N50 greater than 7Mb. The performance could be further improved with more uniform amplification and more accurate sequence alignment. The ability to obtain accurate genome sequences and haplotype information from single cells will enable applications of genome sequencing for diverse clinical needs.

Download Full-text

Tree inference for single-cell data

10.1101/047795 ◽

2016 ◽

Author(s):

Katharina Jahn ◽

Jack Kuipers ◽

Niko Beerenwinkel

Keyword(s):

Single Cell ◽

Search Algorithm ◽

Single Cells ◽

Error Rates ◽

Cancer Therapies ◽

Sequencing Data ◽

Cancer Data ◽

Posterior Probability Distribution ◽

Tree Inference ◽

History Of

AbstractUnderstanding the mutational heterogeneity within tumours is a keystone for the development of efficient cancer therapies. Here, we present SCITE, a stochastic search algorithm to identify the evolutionary history of a tumour from noisy and incomplete mutation profiles of single cells. SCITE comprises a exible MCMC sampling scheme that allows the user to compute the maximum-likelihood mutation history, to sample from the posterior probability distribution, and to estimate the error rates of the underlying sequencing experiments. Evaluation on real cancer data and on simulation studies shows the scalability of SCITE to present-day single-cell sequencing data and improved reconstruction accuracy compared to existing approaches.

Download Full-text

scSLAM-seq reveals core features of transcription dynamics in single cells

10.1101/486852 ◽

2019 ◽

Cited By ~ 2

Author(s):

Florian Erhard ◽

Marisa A.P. Baptista ◽

Tobias Krammer ◽

Thomas Hennig ◽

Marius Lange ◽

...

Keyword(s):

Single Cell ◽

Temporal Dynamics ◽

Single Cells ◽

Transcriptional Response ◽

Gene Promoter ◽

Viral Gene Expression ◽

Viral Gene ◽

Stochastic Nature ◽

Rna Labeling ◽

Induced Changes

AbstractCurrent single-cell RNA sequencing approaches gives a snapshot of a cellular phenotype but convey no information on the temporal dynamics of transcription. Moreover, the stochastic nature of transcription at molecular level is not recovered. Here, we present single-cell SLAM-seq (scSLAM-seq), which integrates metabolic RNA labeling, biochemical nucleoside conversion and single-cell RNA-seq to directly measure total transcript levels and transcriptional activity by differentiating newly synthesized from pre-existing RNA for thousands of genes per single cell. scSLAM-seq recovers the earliest virus-induced changes in cytomegalovirus infection and reveals a so far hidden phase of viral gene expression comprising promiscuous transcription of all kinetic classes. It depicts the stochastic nature of transcription and demonstrates extensive gene-specific differences. These range from stable transcription rates to on-off dynamics which coincide with gene-/promoter-intrinsic features (Tbp-TATA-box interactions and DNA methylation). Gene but not cell-specific features thus explain the heterogeneity in transcriptomes between individual cells and the transcriptional response to perturbations.

Download Full-text

Multiplexed live visualization of cell fate dynamics in hPSCs at single-cell resolution

10.1101/2021.01.30.428961 ◽

2021 ◽

Author(s):

Sungmin Kim ◽

Edward Ren ◽

Paola Marco Casanova ◽

Eugenia Piddini ◽

Rafael Carazo Salas

Keyword(s):

Single Cell ◽

Cell Fate ◽

Fluorescent Proteins ◽

Single Cells ◽

Spatial Location ◽

Time Lapse ◽

Strongly Coupled ◽

Pluripotency Marker ◽

Tightly Coupled ◽

Differentiation Marker

ABSTRACTLive imaging can provide powerful insights into developmental and cellular processes but availability of multiplexable reporters has been limiting. Here we describe ORACLE, a cell fate reporter class in which fluorescent proteins fused with the nucleoporin POM121 are driven by promoters of transcription factors of interest. ORACLE’s nuclear rim localisation therefore enables multiplexing with conventional nuclear reporters. We applied ORACLE to investigate the dynamics of pluripotency exit at single-cell level, using human pluripotent stem cells (hPSCs) imaged by multi-day time-lapse high-content microscopy. Using an ORACLE-OCT4 pluripotency marker we reveal that G1 phase length and OCT4 level are strongly coupled and that spatial location in a colony impacts the timing of pluripotency exit. Combining ORACLE-OCT4 and an ORACLE-SOX1 early neuronal differentiation marker, we visualize in real-time the dynamics of cell fate transition between pluripotency and early neural fate, and show that pluripotency exit and differentiation onset are likely not tightly coupled in single-cells. Thus ORACLE is a powerful tool to enable quantitative studies of spatiotemporal cell fate control.

Download Full-text

pcnaDeep: A Fast and Robust Single-Cell Tracking Method Using Deep-Learning Mediated Cell Cycle Profiling

10.1101/2021.09.19.460933 ◽

2021 ◽

Author(s):

Yifan Gui ◽

Shuang Shuang Xie ◽

Yanan Wang ◽

Ping Wang ◽

Renzhi Yao ◽

...

Keyword(s):

Cell Cycle ◽

Deep Learning ◽

Single Cell ◽

Cell Tracking ◽

Single Cell Analysis ◽

Single Cells ◽

Time Lapse ◽

Molecular Control ◽

A Cell ◽

User Friendly

Motivation: Computational methods that track single-cells and quantify fluorescent biosensors in time-lapse microscopy images have revolutionised our approach in studying the molecular control of cellular decisions. One barrier that limits the adoption of single-cell analysis in biomedical research is the lack of efficient methods to robustly track single-cells over cell division events. Results: Here, we developed an application that automatically tracks and assigns mother-daughter relationships of single-cells. By incorporating cell cycle information from a well-established fluorescent cell cycle reporter, we associate mitosis relationships enabling high fidelity long-term single-cell tracking. This was achieved by integrating a deep-learning based fluorescent PCNA signal instance segmentation module with a cell tracking and cell cycle resolving pipeline. The application offers a user-friendly interface and extensible APIs for customized cell cycle analysis and manual correction for various imaging configurations. Availability and Implementation: pcnaDeep is an open-source Python application under the Apache 2.0 licence. The source code, documentation and tutorials are available at https://github.com/chan-labsite/PCNAdeep.

Download Full-text

Comparing instance segmentation methods for analyzing clonal growth of single cells in microfluidic chips

10.1101/2020.12.31.424955 ◽

2021 ◽

Author(s):

Elliott D. SoRelle ◽

Scott White ◽

Benjamin B. Yellen ◽

Kris C. Wood ◽

Micah A. Luftig ◽

...

Keyword(s):

Single Cell ◽

Cell Biology ◽

Clonal Growth ◽

Single Cells ◽

Time Lapse ◽

Quantitative Information ◽

Cell Segmentation ◽

Morphological Properties ◽

Cell Counts ◽

Segmentation Methods

AbstractAppropriately tailored segmentation techniques can extract detailed quantitative information from biological image datasets to characterize and better understand sample distributions. Practically, high-resolution characterization of biological samples such as cell populations can provide insights into the sources of variance in biomarker expression, drug resistance, and other phenotypic aspects, but it is still unclear what is the best method for extracting this information from large image-based datasets. We present a software pipeline and comparison of multiple image segmentation methods to extract single-cell morphological and fluorescence quantitation from time lapse images of clonal growth rates using a recently reported microfluidic system. The inputs in all pipelines consist of thousands of unprocessed images and the outputs are the detection of cell counts, chamber identifiers, and individual morphological properties of each clone over time detected through multi-channel fluorescence and bright field imaging. Our conclusion is that unsupervised learning methods for cell segmentation substantially outperform supervised statistical methods with respect to accuracy and have key advantages including individual cell instance detection and flexibility through model training. We expect this system and software to have broad utility for researchers interested in high-throughput single-cell biology.

Download Full-text