A high-fidelity cell lineage tracing method for obtaining systematic spatiotemporal gene expression patterns in Caenorhabditis elegans.

Advances in microscopy and fluorescent reporters have allowed us to detect the onset of gene expression on a cell-by-cell basis in a systemic fashion. This information, however, is often encoded in large repositories of images, and developing ways to extract this spatiotemporal expression data is a difficult problem that often uses complex domain-specific methods for each individual data set. We present a more unified approach that incorporates general previous information into a hierarchical probabilistic model to extract spatiotemporal gene expression from 4D confocal microscopy images of developing Caenorhabditis elegans embryos. This approach reduces the overall error rate of our automated lineage tracing pipeline by 3.8-fold, allowing us to routinely follow the C. elegans lineage to later stages of development, where individual neuronal subspecification becomes apparent. Unlike previous methods that often use custom approaches that are organism specific, our method uses generalized linear models and extensions of standard reversible jump Markov chain Monte Carlo methods that can be readily extended to other organisms for a variety of biological inference problems relating to cell fate specification. This modeling approach is flexible and provides tractable avenues for incorporating additional previous information into the model for similar difficult high-fidelity/low error tolerance image analysis problems for systematically applied genomic experiments.

A microfluidic device and computational platform for high-throughput live imaging of gene expression.

To fully describe gene expression dynamics requires the ability to quantitatively capture expression in individual cells over time. Automated systems for acquiring and analyzing real-time images are needed to obtain unbiased data across many samples and conditions. We developed a microfluidics device, the RootArray, in which 64 Arabidopsis thaliana seedlings can be grown and their roots imaged by confocal microscopy over several days without manual intervention. To achieve high throughput, we decoupled acquisition from analysis. In the acquisition phase, we obtain images at low resolution and segment to identify regions of interest. Coordinates are communicated to the microscope to record the regions of interest at high resolution. In the analysis phase, we reconstruct three-dimensional objects from stitched high-resolution images and extract quantitative measurements from a virtual medial section of the root. We tracked hundreds of roots to capture detailed expression patterns of 12 transgenic reporter lines under different conditions.

Automatic annotation of spatial expression patterns via sparse Bayesian factor models.

Advances in reporters for gene expression have made it possible to document and quantify expression patterns in 2D-4D. In contrast to microarrays, which provide data for many genes but averaged and/or at low resolution, images reveal the high spatial dynamics of gene expression. Developing computational methods to compare, annotate, and model gene expression based on images is imperative, considering that available data are rapidly increasing. We have developed a sparse Bayesian factor analysis model in which the observed expression diversity of among a large set of high-dimensional images is modeled by a small number of hidden common factors. We apply this approach on embryonic expression patterns from a Drosophila RNA in situ image database, and show that the automatically inferred factors provide for a meaningful decomposition and represent common co-regulation or biological functions. The low-dimensional set of factor mixing weights is further used as features by a classifier to annotate expression patterns with functional categories. On human-curated annotations, our sparse approach reaches similar or better classification of expression patterns at different developmental stages, when compared to other automatic image annotation methods using thousands of hard-to-interpret features. Our study therefore outlines a general framework for large microscopy data sets, in which both the generative model itself, as well as its application for analysis tasks such as automated annotation, can provide insight into biological questions.

Extraction and comparison of gene expression patterns from 2D RNA in situ hybridization images.

MOTIVATION: Recent advancements in high-throughput imaging have created new large datasets with tens of thousands of gene expression images. Methods for capturing these spatial and/or temporal expression patterns include in situ hybridization or fluorescent reporter constructs or tags, and results are still frequently assessed by subjective qualitative comparisons. In order to deal with available large datasets, fully automated analysis methods must be developed to properly normalize and model spatial expression patterns. RESULTS: We have developed image segmentation and registration methods to identify and extract spatial gene expression patterns from RNA in situ hybridization experiments of Drosophila embryos. These methods allow us to normalize and extract expression information for 78,621 images from 3724 genes across six time stages. The similarity between gene expression patterns is computed using four scoring metrics: mean squared error, Haar wavelet distance, mutual information and spatial mutual information (SMI). We additionally propose a strategy to calculate the significance of the similarity between two expression images, by generating surrogate datasets with similar spatial expression patterns using a Monte Carlo swap sampler. On data from an early development time stage, we show that SMI provides the most biologically relevant metric of comparison, and that our significance testing generalizes metrics to achieve similar performance. We exemplify the application of spatial metrics on the well-known Drosophila segmentation network. AVAILABILITY: A Java webstart application to register and compare patterns, as well as all source code, are available from: http://tools.genome.duke.edu/generegulation/image_analysis/insitu CONTACT: uwe.ohler@duke.edu SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Quantification of transcription factor expression from Arabidopsis images.

MOTIVATION: Confocal microscopy has long provided qualitative information for a variety of applications in molecular biology. Recent advances have led to extensive image datasets, which can now serve as new data sources to obtain quantitative gene expression information. In contrast to microarrays, which usually provide data for many genes at one time point, these image data provide us with expression information for only one gene, but with the advantage of high spatial and/or temporal resolution, which is often lostin microarray samples. RESULTS: We have developed a prototype for the automatic analysis of Arabidopsis confocal images, which show the expression of a single transcription factor by means of GFP reporter constructs. Using techniques from image registration, we are able to address inherent problems of non-rigid transformation and partial mapping, and obtain relative expression values for 13 different tissues in Arabidopsis roots. This provides quantitative information with high spatial resolution, which accurately represents the underlying expression values within the organism. We validate our approach on a data set of 122 images depicting expression patterns of 30 transcription factors, both in terms of registration accuracy, as well as correlation with cell-sorted microarray data. Approaches like this will be useful to lay the groundwork to reconstruct regulatory networks on the level of tissues or even individual cells. AVAILABILITY: Upon request from the authors.