e13028Background: Liquid biopsy has emerged as a powerful, minimally invasive tool for predicting treatment response and survival in breast and other advanced cancers. However, the detection and characterization of circulating tumor cells (CTCs) — a key factor in metastatic progression—remain challenging due to their low frequency and reliance on manual, time-intensive validation using only a couple of established methods for immunofluorescence staining, such as CellSearch. Harnessing deep learning for automated CTC detection and characterization of the blood cells interacting with CTCs holds the potential to advance prognostic evaluations and guide more effective therapies significantly. Methods: Leveraging FDA-approved CellSearch technology and sequencing approaches, we analyzed 2,853 blood specimens, longitudinally collected from 1358 patients with advanced cancer (breast, prostate, etc) and additional diseases. We built a novel deep learning platform, CTCpose, which integrates machine learning and AI-driven image analysis to automate the detection and categorization of CTCs, white blood cells (WBCs), and their clustering interactions. We extracted cellular and nuclear features to enable precise evaluation of individual CTCs, WBCs, homotypic CTC clusters, heterotypic CTC–WBC clusters, and immune cell aggregates. Results: By employing the CTCpose platform, we achieved fully automated identification of CTCs and immune cells, unraveling the spatial organization and functional characteristics of both homotypic and heterotypic clusters. These highly granular assessments revealed clinically significant correlations with patient survival, disease progression, and therapeutic outcomes. Our data underscore the critical role of CTC–immune cell interactions and the dynamic shifts in CTC phenotypes—both as single cells and clusters—in stratifying patients by risk and informing treatment strategies. Conclusions: This work illustrates the transformative power of deep learning in the analysis of liquid biopsy samples. By overcoming the limitations of traditional CTC detection, we have established a robust framework that integrates imaging data with large-scale patient cohorts to deliver predictive models of high clinical relevance. The CTCpose platform not only refines our understanding of CTC–immune cell biology but also paves the way for personalized oncology approaches, highlighting the impactful convergence of artificial intelligence and precision medicine.

Regulation of transcription during embryogenesis is key to development and differentiation. To study transcript expression throughout Caenorhabditis elegans embryogenesis at single-molecule resolution, we developed a high-throughput single-molecule fluorescence in situ hybridization (smFISH) method that relies on computational methods to developmentally stage embryos and quantify individual mRNA molecules in single embryos. We applied our system to sdc-2, a zygotically transcribed gene essential for hermaphrodite development and dosage compensation. We found that sdc-2 is rapidly activated during early embryogenesis by increasing both the number of mRNAs produced per transcription site and the frequency of sites engaged in transcription. Knockdown of sdc-2 and dpy-27, a subunit of the dosage compensation complex (DCC), increased the number of active transcription sites for the X chromosomal gene dpy-23 but not the autosomal gene mdh-1, suggesting that the DCC reduces the frequency of dpy-23 transcription. The temporal resolution from in silico staging of embryos showed that the deletion of a single DCC recruitment element near the dpy-23 gene causes higher dpy-23 mRNA expression after the start of dosage compensation, which could not be resolved using mRNAseq from mixed-stage embryos. In summary, we have established a computational approach to quantify temporal regulation of transcription throughout C. elegans embryogenesis and demonstrated its potential to provide new insights into developmental gene regulation.

Regulation of transcription during embryogenesis is key to development and differentiation. To study transcript expression throughout Caenorhabditis elegans embryogenesis at single-molecule resolution, we developed a high-throughput single-molecule fluorescence in situ hybridization (smFISH) method that relies on computational methods to developmentally stage embryos and quantify individual mRNA molecules in single embryos. We applied our system to sdc-2, a zygotically transcribed gene essential for hermaphrodite development and dosage compensation. We found that sdc-2 is rapidly activated during early embryogenesis by increasing both the number of mRNAs produced per transcription site and the frequency of sites engaged in transcription. Knockdown of sdc-2 and dpy-27, a subunit of the dosage compensation complex (DCC), increased the number of active transcription sites for the X chromosomal gene dpy-23 but not the autosomal gene mdh-1, suggesting that the DCC reduces the frequency of dpy-23 transcription. The temporal resolution from in silico staging of embryos showed that the deletion of a single DCC recruitment element near the dpy-23 gene causes higher dpy-23 mRNA expression after the start of dosage compensation, which could not be resolved using mRNAseq from mixed-stage embryos. In summary, we have established a computational approach to quantify temporal regulation of transcription throughout C. elegans embryogenesis and demonstrated its potential to provide new insights into developmental gene regulation.Competing Interest StatementThe authors have declared no competing interest.

Single-particle electron cryo-microscopy and computational image classification can be used to analyze structural variability in macromolecules and their assemblies. In some cases, a particle may contain different regions that each display a range of distinct conformations. We have developed strategies, implemented within the Frealign and cisTEM image processing packages, to focus classify on specific regions of a particle and detect potential covariance. The strategies are based on masking the region of interest using either a 2-D mask applied to reference projections and particle images, or a 3-D mask applied to the 3-D volume. We show that focused classification approaches can be used to study structural covariance, a concept that is likely to gain more importance as datasets grow in size, allowing the distinction of more structural states and smaller differences between states. Finally, we apply the approaches to an experimental dataset containing the HIV-1 Transactivation Response (TAR) element RNA fused into the large bacterial ribosomal subunit to deconvolve structural mobility within localized regions of interest, and to a dataset containing assembly intermediates of the large subunit to measure structural covariance.

Micro-crystal electron diffraction (MicroED) combines the efficiency of electron scattering with diffraction to allow structure determination from nano-sized crystalline samples in cryoelectron microscopy (cryo-EM). It has been used to solve structures of a diverse set of biomolecules and materials, in some cases to sub-atomic resolution. However, little is known about the damaging effects of the electron beam on samples during such measurements. We assess global and site-specific damage from electron radiation on nanocrystals of proteinase K and of a prion hepta-peptide and find that the dynamics of electron-induced damage follow well-established trends observed in X-ray crystallography. Metal ions are perturbed, disulfide bonds are broken, and acidic side chains are decarboxylated while the diffracted intensities decay exponentially with increasing exposure. A better understanding of radiation damage in MicroED improves our assessment and processing of all types of cryo-EM data.

The ability of fluorescence microscopy to simultaneously image multiple specific molecules of interest has allowed biologists to infer macromolecular organization and colocalization in fixed and live samples. However, a number of factors could affect these analyses, and colocalization is a misnomer. We propose that image similarity coefficient as a better and more descriptive term. In this chapter we will discuss many of the factors involved with determining image similarity including our perception of color in images. In addition, the correct use of several commonly accepted methods such as Pearson’s correlation coefficient, Manders’ overlap coefficient, and Spearman’s ranked correlation coefficient is discussed.

Chromosome inversions are of unique importance in the evolution of genomes and species because when heterozygous with a standard arrangement chromosome, they suppress meiotic crossovers within the inversion. In Drosophila species, heterozygous inversions also cause the interchromosomal effect, whereby the presence of a heterozygous inversion induces a dramatic increase in crossover frequencies in the remainder of the genome within a single meiosis. To date, the interchromosomal effect has been studied exclusively in species that also have high frequencies of inversions in wild populations. We took advantage of a recently developed approach for generating inversions in Drosophila simulans, a species that does not have inversions in wild populations, to ask if there is an interchromosomal effect. We used the existing chromosome 3R balancer and generated a new chromosome 2L balancer to assay for the interchromosomal effect genetically and cytologically. We found no evidence of an interchromosomal effect in D. simulans. To gain insight into the underlying mechanistic reasons, we qualitatively analyzed the relationship between meiotic double-strand break formation and synaptonemal complex assembly. We find that the synaptonemal complex is assembled prior to double-strand break formation as in D. melanogaster; however, we show that the synaptonemal complex is assembled prior to localization of the oocyte determination factor Orb, whereas in D. melanogaster, synaptonemal complex formation does not begin until Orb is localized. Together, our data show heterozygous inversions in D. simulans do not induce an interchromosomal effect and that there are differences in the developmental programming of the early stages of meiosis.

Phosphatidylinositol (PI) is an inositol-containing phospholipid synthesized in the endoplasmic reticulum (ER). PI is a precursor lipid for PI 4,5-bisphosphate (PI(4,5)P) in the plasma membrane (PM) important for Ca signaling in response to extracellular stimuli. Thus, ER-to-PM PI transfer becomes essential for cells to maintain PI(4,5)P homeostasis during receptor stimulation. In this chapter, we discuss two live-cell imaging protocols to analyze ER-to-PM PI transfer at ER-PM junctions, where the two membrane compartments make close appositions accommodating PI transfer. First, we describe how to monitor PI(4,5)P replenishment following receptor stimulation, as a readout of PI transfer, using a PI(4,5)P biosensor and total internal reflection fluorescence microscopy. The second protocol directly visualizes PI transfer proteins that accumulate at ER-PM junctions and mediate PI(4,5)P replenishment with PI in the ER in stimulated cells. These methods provide spatial and temporal analysis of ER-to-PM PI transfer during receptor stimulation and can be adapted to other research questions related to this topic.

New reconstruction techniques are generating connectomes of unprecedented size. These must be analyzed to generate human comprehensible results. The analyses being used fall into three general categories. The first is interactive tools used during reconstruction, to help guide the effort, look for possible errors, identify potential cell classes, and answer other preliminary questions. The second type of analysis is support for formal documents such as papers and theses. Scientific norms here require that the data be archived and accessible, and the analysis reproducible. In contrast to some other "omic" fields such as genomics, where a few specific analyses dominate usage, connectomics is rapidly evolving and the analyses used are often specific to the connectome being analyzed. These analyses are typically performed in a variety of conventional programming language, such as Matlab, R, Python, or C++, and read the connectomic data either from a file or through database queries, neither of which are standardized. In the short term we see no alternative to the use of specific analyses, so the best that can be done is to publish the analysis code, and the interface by which it reads connectomic data. A similar situation exists for archiving connectome data. Each group independently makes their data available, but there is no standardized format and long-term accessibility is neither enforced nor funded. In the long term, as connectomics becomes more common, a natural evolution would be a central facility for storing and querying connectomic data, playing a role similar to the National Center for Biotechnology Information for genomes. The final form of analysis is the import of connectome data into downstream tools such as neural simulation or machine learning. In this process, there are two main problems that need to be addressed. First, the reconstructed circuits contain huge amounts of detail, which must be intelligently reduced to a form the downstream tools can use. Second, much of the data needed for these downstream operations must be obtained by other methods (such as genetic or optical) and must be merged with the extracted connectome.

Automatic image segmentation is critical to scale up electron microscope (EM) connectome reconstruction. To this end, segmentation competitions, such as CREMI and SNEMI, exist to help researchers evaluate segmentation algorithms with the goal of improving them. Because generating ground truth is time-consuming, these competitions often fail to capture the challenges in segmenting larger datasets required in connectomics. More generally, the common metrics for EM image segmentation do not emphasize impact on downstream analysis and are often not very useful for isolating problem areas in the segmentation. For example, they do not capture connectivity information and often over-rate the quality of a segmentation as we demonstrate later. To address these issues, we introduce a novel strategy to enable evaluation of segmentation at large scales both in a supervised setting, where ground truth is available, or an unsupervised setting. To achieve this, we first introduce new metrics more closely aligned with the use of segmentation in downstream analysis and reconstruction. In particular, these include synapse connectivity and completeness metrics that provide both meaningful and intuitive interpretations of segmentation quality as it relates to the preservation of neuron connectivity. Also, we propose measures of segmentation correctness and completeness with respect to the percentage of "orphan" fragments and the concentrations of self-loops formed by segmentation failures, which are helpful in analysis and can be computed without ground truth. The introduction of new metrics intended to be used for practical applications involving large datasets necessitates a scalable software ecosystem, which is a critical contribution of this paper. To this end, we introduce a scalable, flexible software framework that enables integration of several different metrics and provides mechanisms to evaluate and debug differences between segmentations. We also introduce visualization software to help users to consume the various metrics collected. We evaluate our framework on two relatively large public groundtruth datasets providing novel insights on example segmentations.