Search

Abstract Background Transcriptional responses often consist of regulatory modules – sets of genes with a shared expression pattern that are controlled by the same regulatory mechanisms. Previous methods allow dissecting regulatory modules from genomics data, such as expression profiles, protein-DNA binding, and promoter sequences. In cases where physical protein-DNA data are lacking, such methods are essential for the analysis of the underlying regulatory program. Results Here, we present a novel approach for the analysis of modular regulatory programs. Our method – Biochemical Regulatory Network Inference (BRNI) – is based on an algorithm that learns from expression data a biochemically-motivated regulatory program. It describes the expression profiles of gene modules consisting of hundreds of genes using a small number of regulators and affinity parameters. We developed an ensemble learning algorithm that ensures the robustness of the learned model. We then use the topology of the learned regulatory program to guide the discovery of a library of cis-regulatory motifs, and determined the motif compositions associated with each module. We test our method on the cell cycle regulatory program of the fission yeast. We discovered 16 coherent modules, covering diverse processes from cell division to metabolism and associated them with 18 learned regulatory elements, including both known cell-cycle regulatory elements (MCB, Ace2, PCB, ACCCT box) and novel ones, some of which are associated with G2 modules. We integrate the regulatory relations from the expression- and motif-based models into a single network, highlighting specific topologies that result in distinct dynamics of gene expression in the fission yeast cell cycle. Conclusion Our approach provides a biologically-driven, principled way for deconstructing a set of genes into meaningful transcriptional modules and identifying their associated cis-regulatory programs. Our analysis sheds light on the architecture and function of the regulatory network controlling the fission yeast cell cycle, and a similar approach can be applied to the regulatory underpinnings of other modular transcriptional responses.

Embryonic development is highly robust. Morphogenetic variability between embryos (under ideal conditions) is largely quantitative. This robustness stands in contrast to in vitro embryo-like models, which, like most organoids, can display a high degree of tissue morphogenetic variability. The source of this difference is not fully understood. We use the mouse gastruloid model to study the morphogenetic progression of definitive endoderm (DE) and its divergence. We first catalog the different morphologies and characterize their statistics. We then learn predictive models for DE morphotype based on earlier expression and morphology measurements. Finally, we analyze these models to identify key drivers of morphotype variability and devise gastruloid-specific and global interventions that can lower this variability and steer morphotype choice. In the process, we identify two types of coordination lacking in the in vitro model but required for robust gut-tube formation. This approach can help improve the quality and usability of 3D embryo-like models., Competing Interests: Declaration of interests The authors declare no competing interests., (Copyright © 2024 Elsevier Inc. All rights reserved.)

The cultivation of meat using in vitro grown animal stem cells offers a promising solution to pressing global concerns around climate change, ethical considerations, and public health. However, cultivated meat introduces an unprecedented necessity: the generation of mass scales of cellular biomaterial, achieved by fostering cell proliferation within bioreactors. Existing methods for in vitro cell proliferation encounter substantial challenges in terms of both scalability and economic viability. Within this perspective, we discuss the current landscape of cell proliferation optimization, focusing on approaches pertinent to cellular agriculture. We examine the mechanisms governing proliferation rates, while also addressing intrinsic and conditional rate limitations. Furthermore, we expound upon prospective strategies that could lead to a significant enhancement of the overall scalability and cost-efficiency of the cell proliferation phase within the cultivated meat production process. By exploring knowledge from basic cell cycle studies, pathological contexts and tissue engineering, we may identify innovative solutions toward optimizing cell expansion., Competing Interests: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest., (Copyright © 2024 Hauser, Zirman, Rak and Nachman.)

Motivation: Technical differences between gene expression sequencing experiments can cause variations in the data in the form of batch effect biases. These do not represent true biological variations between samples and can lead to false conclusions or hinder the ability to integrate multiple datasets. Since there is a growing need for the joint analysis of single - cell sequencing datasets from different sources, there is also a need to correct the resulting batch effects while maintaining the true biological variations in the data., Results: We developed a semi-supervised deep learning architecture called Autoencoder-based Batch Correction (ABC) for integrating single - cell sequencing datasets. Our method removes batch effects through a guided process of data compression using supervised cell type classifier branches for biological signal retention. It aligns the different batches using an adversarial training approach. We comprehensively evaluate the performance of our method using four single - cell sequencing datasets and multiple measures for batch effect removal and biological variation conservation. ABC outperforms 10 state-of-the-art methods for this task including Seurat, scGen, ComBat, scanorama, scVI, scANVI, AutoClass, Harmony, scDREAMER, and CLEAR, correcting various types of batch effects while preserving intricate biological variations., Competing Interests: None declared., (© The Author(s) 2023. Published by Oxford University Press.)

The young field of gastruloids brings promise to modeling and understanding early embryonic development. However, being a complex model, gastruloids are prone to variability at different levels. In this perspective, we define the different levels of gastruloid variability, and parameters over which it can be measured. We discuss potential sources for variability, and then propose methods to better control and reduce it. We provide an example from definitive endoderm progression in gastruloids, where we harness gastruloid-to-gastruloid variation in early parameters to identify key driving factors for endoderm morphology. We then devise interventions that steer morphological outcome. A better control over the developmental progression of gastruloids will enhance their utility in both basic research and biomedical applications., (© 2023 The Author(s).)

During embryogenesis, organisms acquire their shape given boundary conditions that impose geometrical, mechanical and biochemical constraints. A detailed integrative understanding how these morphogenetic information modules pattern and shape the mammalian embryo is still lacking, mostly owing to the inaccessibility of the embryo in vivo for direct observation and manipulation. These impediments are circumvented by the developmental engineering of embryo-like structures (stembryos) from pluripotent stem cells that are easy to access, track, manipulate and scale. Here, we explain how unlocking distinct levels of embryo-like architecture through controlled modulations of the cellular environment enables the identification of minimal sets of mechanical and biochemical inputs necessary to pattern and shape the mammalian embryo. We detail how this can be complemented with precise measurements and manipulations of tissue biochemistry, mechanics and geometry across spatial and temporal scales to provide insights into the mechanochemical feedback loops governing embryo morphogenesis. Finally, we discuss how, even in the absence of active manipulations, stembryos display intrinsic phenotypic variability that can be leveraged to define the constraints that ensure reproducible morphogenesis in vivo., Competing Interests: Competing interests The authors declare no competing or financial interests., (© 2021. Published by The Company of Biologists Ltd.)

The segregation of definitive endoderm (DE) from bipotent mesendoderm progenitors leads to the formation of two distinct germ layers. Dissecting DE commitment and onset has been challenging as it occurs within a narrow spatiotemporal window in the embryo. Here, we employ a dual Bra/Sox17 reporter cell line to study DE onset dynamics. We find Sox17 expression initiates in vivo in isolated cells within a temporally restricted window. In 2D and 3D in vitro models, DE cells emerge from mesendoderm progenitors at a temporally regular, but spatially stochastic pattern, which is subsequently arranged by self-sorting of Sox17 + cells. A subpopulation of Bra-high cells commits to a Sox17+ fate independent of external Wnt signal. Self-sorting coincides with upregulation of E-cadherin but is not necessary for DE differentiation or proliferation. Our in vivo and in vitro results highlight basic rules governing DE onset and patterning through the commonalities and differences between these systems., Competing Interests: The authors declare no competing interests., (© 2021 The Authors.)

Tungsten disulfide nanotubes (WS 2 -NTs) were found to be very active for photothermal therapy. However, their lack of stability in aqueous solutions inhibits their use in many applications, especially in biomedicine. Few attempts were made to chemically functionalize the surface of the NTs to improve their dispersability. Here, we present a new polymerization method using cerium-doped maghemite nanoparticles (CM-NPs) as magnetic nanosized linkers between the WS 2 -NT surface and pyrrole-N-propionic acid monomers, which allow in situ polymerization onto the composite surface. This unique composite is magnetic, and contains two active entities for photothermal therapy-WS 2 and the polypyrrole. The photothermal activity of the composite was tested at a wavelength of 808 nm, and significant thermal activity was observed. Moreover, the polycarboxylated polymeric coating of the NTs enables effective linkage of additional molecules or drugs via covalent bonding. In addition, a new method was established for large-scale synthesis of CM-NPs and WS 2 -NT-CM composites., (© 2021. The Author(s).)

During early embryogenesis, mechanical constraints and localized biochemical signals co-occur around anteroposterior axis determination and symmetry breaking. Their relative roles, however, are hard to tease apart in vivo Using brachyury (Bra), a primitive streak and mesendoderm marker in mouse embryoid bodies (EBs), we studied how contact, biochemical cues and neighboring cell cues affect the positioning of a primitive streak-like locus and thus determine the anteroposterior axis. We show that a Bra-competent layer must be formed in the EB before Bra expression initiates, and that Bra onset locus position is biased by contact points of the EB with its surrounding, probably through modulation of chemical cues rather than by mechanical signaling. We can push or pull Bra onset away from contact points by introducing a separate localized Wnt signal source, or maneuver Bra onset to a few loci or to an isotropic peripheral pattern. Furthermore, we show that Foxa2-positive cells are predictive of the future location of Bra onset, demonstrating an earlier symmetry-breaking event. Our analysis of factors affecting symmetry breaking and spatial fate choice during this developmental process could prove valuable for in vitro differentiation and organoid formation., Competing Interests: Competing interestsThe authors declare no competing or financial interests., (© 2019. Published by The Company of Biologists Ltd.)

Ectopic transcription factor expression enables reprogramming of somatic cells to pluripotency, albeit with generally low efficiency. Despite steady progress in the field, the exact molecular mechanisms that coordinate this remarkable transition still remain largely elusive. To better characterize the final steps of pluripotency induction, we optimized an experimental system where pluripotent stem cells are differentiated for set intervals before being reintroduced to pluripotency-supporting conditions. Using this approach, we identify a transient period of high-efficiency reprogramming where ectopic transcription factors, but not serum/LIF alone, rapidly revert cells to pluripotency with near 100% efficiency. After this period, cells reprogram with somatic-like kinetics and efficiencies. We identify a set of OCT4 bound cis-regulatory elements that are dynamically regulated during this transient phase and appear central to facilitating reprogramming. Interestingly, these regions remain hypomethylated during in vitro and in vivo differentiation, which may allow them to act as primary targets of ectopically induced factors during somatic cell reprogramming.

The blastocyst stage and the subsequent implantation are critical for a successful pregnancy, yet are challenging to study in vivo. In this issue of Stem Cell Reports, Kime et al. (2019) describe a novel way to generate blastocyst-like structures only from pluripotent stem cells. These structures mimic several aspects of the early embryo, offering a new promising tool to study this stage., (Copyright © 2019 The Authors. Published by Elsevier Inc. All rights reserved.)

Ubiquitin (Ub) receptors respond to ubiquitylation signals. They bind ubiquitylated substrates and exert their activity in situ. Intriguingly, Ub receptors themselves undergo rapid ubiquitylation and deubiquitylation. Here we asked what is the function of ubiquitylation of Ub receptors? We focused on yeast epsin, a Ub receptor that decodes the ubiquitylation signal of plasma membrane proteins into an endocytosis response. Using mass spectrometry, we identified lysine-3 as the major ubiquitylation site in the epsin plasma membrane binding domain. By projecting this ubiquitylation site onto our crystal structure, we hypothesized that this modification would compete with phosphatidylinositol-4,5-bisphosphate (PIP 2 ) binding and dissociate epsin from the membrane. Using an E. coli -based expression of an authentic ubiquitylation apparatus, we purified ubiquitylated epsin. We demonstrated in vitro that in contrast to apo epsin, the ubiquitylated epsin does not bind to either immobilized PIPs or PIP 2 -enriched liposomes. To test this hypothesis in vivo, we mimicked ubiquitylation by the fusion of Ub at the ubiquitylation site. Live cell imaging demonstrated that the mimicked ubiquitylated epsin dissociates from the membrane. Our findings suggest that ubiquitylation of the Ub receptors dissociates them from their products to allow binding to a new ubiquitylated substrates, consequently promoting cyclic activity of the Ub receptors.

Nanostructures of transition-metal dichalcogenides (TMDC) have raised scientific interest in the last few decades. Tungsten disulfide (WS 2 ) nanotubes and nanoparticles are among the most extensively studied members in this group, and are used for, e.g., polymer reinforcement, lubrication and electronic devices. Their biocompatibility and low toxicity make them suitable for medical and biological applications. One potential application is photothermal therapy (PTT), a method for the targeted treatment of cancer, in which a light-responsive material is irradiated with a laser in the near-infrared range. In the current article we present WS 2 nanotubes functionalized with previously reported ceric ammonium nitrate-maghemite (CAN-mag) nanoparticles, used for PTT. Functionalization of the nanotubes with CAN-mag nanoparticles resulted in a magnetic nanocomposite. When tested in vitro with two types of cancer cells, the functionalized nanotubes showed a better PTT activity compared to non-functionalized nanotubes, as well as reduced aggregation and the ability to add a second-step functionality. This ability is demonstrated here with two polymers grafted onto the nanocomposite surface, and other functionalities could be additional cancer therapy agents for achieving increased therapeutic activity.

In experimental evolution, scientists evolve organisms in the lab, typically by challenging them to new environmental conditions. How best to evolve a desired trait? Should the challenge be applied abruptly, gradually, periodically, sporadically? Should one apply chemical mutagenesis, and do strains with high innate mutation rate evolve faster? What are ideal population sizes of evolving populations? There are endless strategies, beyond those that can be exposed by individual labs. We therefore arranged a community challenge, Evolthon, in which students and scientists from different labs were asked to evolve Escherichia coli or Saccharomyces cerevisiae for an abiotic stress-low temperature. About 30 participants from around the world explored diverse environmental and genetic regimes of evolution. After a period of evolution in each lab, all strains of each species were competed with one another. In yeast, the most successful strategies were those that used mating, underscoring the importance of sex in evolution. In bacteria, the fittest strain used a strategy based on exploration of different mutation rates. Different strategies displayed variable levels of performance and stability across additional challenges and conditions. This study therefore uncovers principles of effective experimental evolutionary regimens and might prove useful also for biotechnological developments of new strains and for understanding natural strategies in evolutionary arms races between species. Evolthon constitutes a model for community-based scientific exploration that encourages creativity and cooperation., Competing Interests: The authors have declared that no competing interests exist.

Multifunctional nanoparticles have attracted significant interest as biomedical vehicles, combining diagnostic, imaging, and therapeutic properties. We describe herein the construction of new nanoparticle conjugates comprising WS 2 nanorods (NRs) coupled to fluorescent carbon dots (C-dots). We show that the WS 2 -C-dot hybrids integrate the unique physical properties of the two species, specifically the photothermal activity of the WS 2 NRs upon irradiation with near-infrared (NIR) light and the excitation-dependent luminescence emission of the C-dots. The WS 2 -C-dot NRs have been shown to be non-cytotoxic and have been successfully employed for multicolour cell imaging and targeted cell killing under NIR irradiation, pointing to their potential utilization as effective therapeutic vehicles., (© 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.)

Embryonic stem cells can spontaneously differentiate into cell types of all germ layers within embryoid bodies (EBs) in a highly variable manner. Whether there exists an intrinsic differentiation program common to all EBs is unknown. Here, we present a novel combination of high-throughput live two-photon imaging and gene expression profiling to study early differentiation dynamics spontaneously occurring within developing EBs. Onset timing of Brachyury-GFP was highly variable across EBs, while the spatial patterns as well as the dynamics of mesendodermal progression following onset were remarkably similar. We therefore defined a 'developmental clock' using the Brachyury-GFP signal onset timing. Mapping snapshot gene expression measurements to this clock revealed their temporal trends, indicating that loss of pluripotency, formation of primitive streak and mesodermal lineage progression are synchronized in EBs. Exogenous activation of Wnt or BMP signaling accelerated the intrinsic clock. CHIR down-regulated Wnt3, allowing insights into dependency mechanisms between canonical Wnt signaling and multiple genes. Our findings reveal a developmental clock characteristic of an early differentiation program common to all EBs, further establishing them as an in vitro developmental model.

Transferring Saccharomyces cerevisiae cells to water is known to extend their lifespan. However, it is unclear whether this lifespan extension is due to slowing the aging process or merely keeping old yeast alive. Here we show that in water-transferred yeast, the toxicity of polyQ proteins is decreased and the aging biomarker 47Q aggregates at a reduced rate and to a lesser extent. These beneficial effects of water-transfer could not be reproduced by diluting the growth medium and depended on de novo protein synthesis and proteasomes levels. Interestingly, we found that upon water-transfer 27 proteins are downregulated, 4 proteins are upregulated and 81 proteins change their intracellular localization, hinting at an active genetic program enabling the lifespan extension. Furthermore, the aging-related deterioration of the heat shock response (HSR), the unfolded protein response (UPR) and the endoplasmic reticulum-associated protein degradation (ERAD), was largely prevented in water-transferred yeast, as the activities of these proteostatic network pathways remained nearly as robust as in young yeast. The characteristics of young yeast that are actively maintained upon water-transfer indicate that the extended lifespan is the outcome of slowing the rate of the aging process.

Developmental processes in cells require a series of complex steps. Often only a single master regulator activates genes in these different steps. This poses several challenges: some targets need to be ordered temporally, while co-functional targets may need to be synchronized in both time and expression level. Here we study in single cells the dynamic activation patterns of early meiosis genes in budding yeast, targets of the meiosis master regulator Ime1. We quantify the individual roles of the promoter and protein levels in expression pattern control, as well as the roles of individual promoter elements. We find a consistent expression pattern difference between a non-cofunctional pair of genes, and a highly synchronized activation of a co-functional pair. We show that dynamic control leading to these patterns is distributed between promoter, gene and external regions. Through specific reciprocal changes to the promoters of pairs of genes, we show that different genes can use different promoter elements to reach near identical activation patterns.

Reprogramming to pluripotency is a low-efficiency process at the population level. Despite notable advances to molecularly characterize key steps, several fundamental aspects remain poorly understood, including when the potential to reprogram is first established. Here, we apply live-cell imaging combined with a novel statistical approach to infer when somatic cells become fated to generate downstream pluripotent progeny. By tracing cell lineages from several divisions before factor induction through to pluripotent colony formation, we find that pre-induction sister cells acquire similar outcomes. Namely, if one daughter cell contributes to a lineage that generates induced pluripotent stem cells (iPSCs), its paired sibling will as well. This result suggests that the potential to reprogram is predetermined within a select subpopulation of cells and heritable, at least over the short term. We also find that expanding cells over several divisions prior to factor induction does not increase the per-lineage likelihood of successful reprogramming, nor is reprogramming fate correlated to neighboring cell identity or cell-specific reprogramming factor levels. By perturbing the epigenetic state of somatic populations with Ezh2 inhibitors prior to factor induction, we successfully modulate the fraction of iPSC-forming lineages. Our results therefore suggest that reprogramming potential may in part reflect preexisting epigenetic heterogeneity that can be tuned to alter the cellular response to factor induction., (© 2015 The Authors.)

The timing of a cellular event often hides critical information on the process leading to the event. Our ability to measure event times in single cells along with other quantities allow us to learn about the drivers of the timed process and its downstream effects. In this review, we cover different types of events that have been timed in single cells, methods to time such events and types of analysis that have been applied to event timings. We show how different timing distributions suggest different natures for the process. The statistical relations between the timing of different events may reveal how their respective processes are related biologically: Do they occur in sequence or in parallel? Are they independent or inter-dependent? Finally, quantifying morphological and molecular variables may help assess their contribution to the timing of an event and its related process.

The study of cell responses to environmental changes poses many experimental challenges: cells need to be imaged under changing conditions, often in a comparative manner. Multiwell plates are routinely used to compare many different strains or cell lines, but allow limited control over the environment dynamics. Microfluidic devices, on the other hand, allow exquisite dynamic control over the surrounding conditions, but it is challenging to image and distinguish more than a few strains in them. Here we describe a method to easily and rapidly manufacture a microfluidic device capable of applying dynamically changing conditions to multiple distinct yeast strains in one channel. The device is designed and manufactured by simple means without the need for soft lithography. It is composed of a Y-shaped flow channel attached to a second layer harboring microwells. The strains are placed in separate microwells, and imaged under the exact same dynamic conditions. We demonstrate the use of the device for measuring protein localization responses to pulses of nutrient changes in different yeast strains.

The Gram-negative bacterium Pseudomonas syringae pv. tomato (Pst) is the causal agent of speck disease in tomato. Pst pathogenicity depends on a type III secretion system that delivers effector proteins into host cells, where they promote disease by manipulating processes to the advantage of the pathogen. Previous studies identified seven Pst effectors that inhibit growth when expressed in yeast under normal growth conditions, suggesting that they interfere with cellular processes conserved in yeast and plants. We hypothesized that effectors also target conserved cellular processes that are required for yeast growth only under stress conditions. We therefore examined phenotypes induced by expression of Pst effectors in yeast grown in the presence of various stressors. Out of 29 effectors tested, five (HopX1, HopG1, HopT1-1, HopH1 and AvrPtoB) displayed growth inhibition phenotypes only in combination with stress conditions. Viability assays revealed that the HopX1 effector caused loss of cell viability under prolonged osmotic stress. Using transcription reporters, we found that HopX1 attenuated the activation of the high osmolarity glycerol (HOG) mitogen-activated protein kinase (MAPK) pathway, which is responsible for yeast survival under osmotic stress, while other MAPK pathways were mildly affected by HopX1. Interestingly, HopX1-mediated phenotypes in yeast were dependent on the putative transglutaminase catalytic triad of the effector. This study enlarges the pool of phenotypes available for the functional analysis of Pst type III effectors in yeast, and exemplifies how analysis of phenotypes detected in yeast under stress conditions can lead to the identification of eukaryotic cellular processes affected by bacterial effectors.

Aging-related neurodegenerative disorders, such as Parkinson's, Alzheimer's and Huntington's diseases, are characterized by accumulation of protein aggregates in distinct neuronal cells that eventually die. In Huntington's disease, the protein huntingtin forms aggregates, and the age of disease onset is inversely correlated to the length of the protein's poly-glutamine tract. Using quantitative assays to estimate microscopically and capture biochemically protein aggregates, here we study in Saccharomyces cerevisiae aging-related aggregation of GFP-tagged, huntingtin-derived proteins with different polyQ lengths. We find that the short 25Q protein never aggregates whereas the long 103Q version always aggregates. However, the mid-size 47Q protein is soluble in young logarithmically growing yeast but aggregates as the yeast cells enter the stationary phase and age, allowing us to plot an "aggregation timeline". This aging-dependent aggregation was associated with increased cytotoxicity. We also show that two aging-related genes, SIR2 and HSF1, affect aggregation of the polyQ proteins. In Δsir2 strain the aging-dependent aggregation of the 47Q protein is aggravated, while overexpression of the transcription factor Hsf1 attenuates aggregation. Thus, the mid-size 47Q protein and our quantitative aggregation assays provide valuable tools to unravel the roles of genes and environmental conditions that affect aging-related aggregation.

The study of induced pluripotency often relies on experimental approaches that average measurements across a large population of cells, the majority of which do not become pluripotent. Here we used high-resolution, time-lapse imaging to trace the reprogramming process over 2 weeks from single mouse embryonic fibroblasts (MEFs) to pluripotency factor-positive colonies. This enabled us to calculate a normalized cell-of-origin reprogramming efficiency that takes into account only the initial MEFs that respond to form reprogrammed colonies rather than the larger number of final colonies. Furthermore, this retrospective analysis revealed that successfully reprogramming cells undergo a rapid shift in their proliferative rate that coincides with a reduction in cellular area. This event occurs as early as the first cell division and with similar kinetics in all cells that form induced pluripotent stem (iPS) cell colonies. These data contribute to the theoretical modeling of reprogramming and suggest that certain parts of the reprogramming process follow defined rather than stochastic steps.

Cell-to-cell variability in the timing of cell-fate changes can be advantageous for a population of single-celled organisms growing in a fluctuating environment. We study timing variability during meiosis in Saccharomyces cerevisiae, initiated upon nutritional starvation. We use time-lapse fluorescence microscopy to measure the timing of meiotic events in single cells and find that the duration of meiosis is highly variable between cells. This variability is concentrated between the beginning of starvation and the onset of early meiosis genes. Cell-cycle variability and nutritional history have little effect on this timing variability. Rather, variation in the production rate of the meiotic master regulator Ime1 and its gradual increase over time govern this variability, and cell size effects are channeled through Ime1. These results tie phenotypic variability with expression dynamics of a transcriptional regulator and provide a general framework for the study of temporal developmental processes.

Constantly improving gene expression profiling technologies are expected to provide understanding and insight into cancer-related cellular processes. Gene expression data is also expected to significantly aid in the development of efficient cancer diagnosis and classification platforms. In this work we examine three sets of gene expression data measured across sets of tumor(s) and normal clinical samples: The first set consists of 2,000 genes, measured in 62 epithelial colon samples (Alon et al., 1999). The second consists of approximately equal to 100,000 clones, measured in 32 ovarian samples (unpublished extension of data set described in Schummer et al. (1999)). The third set consists of approximately equal to 7,100 genes, measured in 72 bone marrow and peripheral blood samples (Golub et al, 1999). We examine the use of scoring methods, measuring separation of tissue type (e.g., tumors from normals) using individual gene expression levels. These are then coupled with high-dimensional classification methods to assess the classification power of complete expression profiles. We present results of performing leave-one-out cross validation (LOOCV) experiments on the three data sets, employing nearest neighbor classifier, SVM (Cortes and Vapnik, 1995), AdaBoost (Freund and Schapire, 1997) and a novel clustering-based classification technique. As tumor samples can differ from normal samples in their cell-type composition, we also perform LOOCV experiments using appropriately modified sets of genes, attempting to eliminate the resulting bias. We demonstrate success rate of at least 90% in tumor versus normal classification, using sets of selected genes, with, as well as without, cellular-contamination-related members. These results are insensitive to the exact selection mechanism, over a certain range.

DNA hybridization arrays simultaneously measure the expression level for thousands of genes. These measurements provide a "snapshot" of transcription levels within the cell. A major challenge in computational biology is to uncover, from such measurements, gene/protein interactions and key biological features of cellular systems. In this paper, we propose a new framework for discovering interactions between genes based on multiple expression measurements. This framework builds on the use of Bayesian networks for representing statistical dependencies. A Bayesian network is a graph-based model of joint multivariate probability distributions that captures properties of conditional independence between variables. Such models are attractive for their ability to describe complex stochastic processes and because they provide a clear methodology for learning from (noisy) observations. We start by showing how Bayesian networks can describe interactions between genes. We then describe a method for recovering gene interactions from microarray data using tools for learning Bayesian networks. Finally, we demonstrate this method on the S. cerevisiae cell-cycle measurements of Spellman et al. (1998).