Correlation Analysis of Histopathology and Proteogenomics Data for Breast Cancer.

Tumors are heterogeneous tissues with different types of cells such as cancer cells, fibroblasts, and lymphocytes. Although the morphological features of tumors are critical for cancer diagnosis and prognosis, the underlying molecular events and genes for tumor morphology are far from being clear. With the advancement in computational pathology and accumulation of large amount of cancer samples with matched molecular and histopathology data, researchers can carry out integrative analysis to investigate this issue. In this study, we systematically examine the relationships between morphological features and various molecular data in breast cancers. Specifically, we identified 73 breast cancer patients from the TCGA and CPTAC projects matched whole slide images, RNA-seq, and proteomic data. By calculating 100 different morphological features and correlating them with the transcriptomic and proteomic data, we inferred four major biological processes associated with various interpretable morphological features. These processes include metabolism, cell cycle, immune response, and extracellular matrix development, which are all hallmarks of cancers and the associated morphological features are related to area, density, and shapes of epithelial cells, fibroblasts, and lymphocytes. In addition, protein specific biological processes were inferred solely from proteomic data, suggesting the importance of proteomic data in obtaining a holistic understanding of the molecular basis for tumor tissue morphology. Furthermore, survival analysis yielded specific morphological features related to patient prognosis, which have a strong association with important molecular events based on our analysis. Overall, our study demonstrated the power for integrating multiple types of biological data for cancer samples in generating new hypothesis as well as identifying potential biomarkers predicting patient outcome. Future work includes causal analysis to identify key regulators for cancer tissue development and validating the findings using more independent data sets.

Body and Wing Allometries Reveal Flight-Fecundity Tradeoff in Response to Larval Provisioning in Osmia lignaria (Hymenoptera: Megachilidae).

Variation in body size has important implications for physical performance and fitness. For insects, adult size and morphology are determined by larval growth and metamorphosis. Female blue orchard bees, Osmia lignaria, (Say) provision a finite quantity of food to their offspring. In this study, we asked how provision-dependent variation in size changes adult morphology. We performed a diet manipulation in which some larvae were starved in the final instar and some were given unlimited food. We examined the consequences on adult morphology in two ways. First, allometric relationships between major body regions (head, thorax, abdomen) and total body mass were measured to determine relative growth of these structures. Second, morphometrics that are critical for flight (wing area, wing loading, and extra flight power index) were quantified. Head and thorax mass had hyperallometric relationships with body size, indicating these parts become disproportionately large in adults when larvae are given copious provisions. However, abdominal mass and wing area increased hypoallometrically with body size. Thus, large adults had disproportionately lighter abdomens and smaller wing areas than smaller adults. Though both males and females followed these general patterns, allometric patterns were affected by sex. For flight metrics, small adults had reduced wing loading and an increased extra flight power index. These results suggest that diet quantity alters development in ways that affect the morphometric trait relationships in adult O. lignaria and may lead to functional differences in performance.

Metamorphosis is induced by food absence rather than a critical weight in the solitary bee, Osmia lignaria.

Body size is an important phenotypic trait that correlates with performance and fitness. For determinate growing insects, body size variation is determined by growth rate and the mechanisms that stop growth at the end of juvenile growth. Endocrine mechanisms regulate growth cessation, and their relative timing along development shapes phenotypic variation in body size and development time. Larval insects are generally hypothesized to initiate metamorphosis once they attain a critical weight. However, the mechanisms underlying the critical weight have not been resolved even for well-studied insect species. More importantly, critical weights may or may not be generalizable across species. In this study, we characterized the developmental aspects of size regulation in the solitary bee, Osmia lignaria We demonstrate that starvation cues metamorphosis in O. lignaria and that a critical weight does not exist in this species. Larvae initiated pupation <24 h after food was absent. However, even larvae fed ad libitum eventually underwent metamorphosis, suggesting that some secondary mechanism regulates metamorphosis when provisions are not completely consumed. We show that metamorphosis could be induced by precocene treatment in the presence of food, which suggests that this decision is regulated through juvenile hormone signaling. Removing food at different larval masses produced a 10-fold difference in mass between smallest and largest adults. We discuss the implications of body size variation for insect species that are provided with a fixed quantity of provisions, including many bees which have economic value as pollinators.

Phenotypic integration in an extended phenotype: among-individual variation in nest-building traits of the alfalfa leafcutting bee (Megachile rotundata).

Structures such as nests and burrows are an essential component of many organisms' life-cycle and require a complex sequence of behaviours. Because behaviours can vary consistently among individuals and be correlated with one another, we hypothesized that these structures would (1) show evidence of among-individual variation, (2) be organized into distinct functional modules and (3) show evidence of trade-offs among functional modules due to limits on energy budgets. We tested these hypotheses using the alfalfa leafcutting bee, Megachile rotundata, a solitary bee and important crop pollinator. Megachile rotundata constructs complex nests by gathering leaf materials to form a linear series of cells in pre-existing cavities. In this study, we examined variation in the following nest construction traits: reproduction (number of cells per nest and nest length), nest protection (cap length and number of leaves per cap), cell construction (cell size and number of leaves per cell) and cell provisioning (cell mass) from 60 nests. We found a general decline in investment in cell construction and provisioning with each new cell built. In addition, we found evidence for both repeatability and plasticity in cell provisioning with little evidence for trade-offs among traits. Instead, most traits were positively, albeit weakly, correlated (r ~ 0.15), and traits were loosely organized into covarying modules. Our results show that individual differences in nest construction are detectable at a level similar to that of other behavioural traits and that these traits are only weakly integrated. This suggests that nest components are capable of independent evolutionary trajectories.

Comparison of the Chemical Compositions of the Cuticle and Dufour's Gland of Two Solitary Bee Species from Laboratory and Field Conditions.

Species-specific biochemistry, morphology, and function of the Dufour's gland have been investigated for social bees and some non-social bee families. Most of the solitary bees previously examined are ground-nesting bees that use Dufour's gland secretions to line brood chambers. This study examines the chemistry of the cuticle and Dufour's gland of cavity-nesting Megachile rotundata and Osmia lignaria, which are species managed for crop pollination. Glandular and cuticular lipid compositions were characterized and compared to each other and according to the nesting experience of adult females. Major lipid classes found were hydrocarbons, free fatty acids, and wax esters. Many components were common to the cuticle and Dufour's glands of each species, yet not identical in number or relative composition. Wax esters and fatty acids were more prevalent in Dufour's glands of M. rotundata than on cuticles. Wax esters were more abundant on cuticles of O. lignaria than in Dufour's glands. In both species, fatty acids were more prevalent in glands of field-collected females compared to any other sample type. Chemical profiles of cuticles and glands were distinct from each other, and, for O. lignaria, profiles of laboratory-maintained bees could be distinguished from those of field-collected bees. Comparison of percentiles of individual components of cuticular and glandular profiles of the same bee showed that the proportions of some cuticular components were predictive of the proportion of the same glandular components, especially for nesting females. Lastly, evidence suggested that Dufour's gland is the major source of nest-marking substances in M. rotundata, but evidence for this role in O. lignaria was less conclusive.

Mass and volume growth of an insect tracheal system within a single instar.

Organisms must accommodate oxygen delivery to developing tissues as body mass increases during growth. In insects, the growth of the respiratory system has been assumed to occur only during molts, whereas body mass and volume increase during the larval stages between molts. This decouples whole-body growth from the growth of the oxygen supply system. This assumption is derived from the observation that the insect respiratory system is an invagination of the exoskeleton, which must be shed during molts for continued growth to occur. Here, we provide evidence that this assumption is incorrect. We found that the respiratory system increases substantially in both mass and volume within the last larval instar of Manduca sexta larvae, and that the growth of the respiratory system changes with diet quality, potentially as a consequence of shifting metabolic demands.

Percent lipid is associated with body size but not task in the bumble bee Bombus impatiens.

In some group-living organisms, labor is divided among individuals. This allocation to particular tasks is frequently stable and predicted by individual physiology. Social insects are excellent model organisms in which to investigate the interplay between physiology and individual behavior, as division of labor is an important feature within colonies, and individual physiology varies among the highly related individuals of the colony. Previous studies have investigated what factors are important in determining how likely an individual is, compared to nestmates, to perform certain tasks. One such task is foraging. Corpulence (i.e., percent lipid) has been shown to determine foraging propensity in honey bees and ants, with leaner individuals being more likely to be foragers. Is this a general trend across all social insects? Here we report data analyzing the individual physiology, specifically the percent lipid, of worker bumble bees (Bombus impatiens) from whom we also analyze behavioral task data. Bumble bees are also unusual among the social bees in that workers may vary widely in size. Surprisingly we find that, unlike other social insects, percent lipid is not associated with task propensity. Rather, body size closely predicts individual relative lipid stores, with smaller worker bees being allometrically fatter than larger worker bees.

Spatial cell type composition in normal and Alzheimers human brains is revealed using integrated mouse and human single cell RNA sequencing.

Single-cell RNA sequencing (scRNA-seq) resolves heterogenous cell populations in tissues and helps to reveal single-cell level function and dynamics. In neuroscience, the rarity of brain tissue is the bottleneck for such study. Evidence shows that, mouse and human share similar cell type gene markers. We hypothesized that the scRNA-seq data of mouse brain tissue can be used to complete human data to infer cell type composition in human samples. Here, we supplement cell type information of human scRNA-seq data, with mouse. The resulted data were used to infer the spatial cellular composition of 3702 human brain samples from Allen Human Brain Atlas. We then mapped the cell types back to corresponding brain regions. Most cell types were localized to the correct regions. We also compare the mapping results to those derived from neuronal nuclei locations. They were consistent after accounting for changes in neural connectivity between regions. Furthermore, we applied this approach on Alzheimer's brain data and successfully captured cell pattern changes in AD brains. We believe this integrative approach can solve the sample rarity issue in the neuroscience.

Deep learning-based cancer survival prognosis from RNA-seq data: approaches and evaluations.

BACKGROUND: Recent advances in kernel-based Deep Learning models have introduced a new era in medical research. Originally designed for pattern recognition and image processing, Deep Learning models are now applied to survival prognosis of cancer patients. Specifically, Deep Learning versions of the Cox proportional hazards models are trained with transcriptomic data to predict survival outcomes in cancer patients. METHODS: In this study, a broad analysis was performed on TCGA cancers using a variety of Deep Learning-based models, including Cox-nnet, DeepSurv, and a method proposed by our group named AECOX (AutoEncoder with Cox regression network). Concordance index and p-value of the log-rank test are used to evaluate the model performances. RESULTS: All models show competitive results across 12 cancer types. The last hidden layers of the Deep Learning approaches are lower dimensional representations of the input data that can be used for feature reduction and visualization. Furthermore, the prognosis performances reveal a negative correlation between model accuracy, overall survival time statistics, and tumor mutation burden (TMB), suggesting an association among overall survival time, TMB, and prognosis prediction accuracy. CONCLUSIONS: Deep Learning based algorithms demonstrate superior performances than traditional machine learning based models. The cancer prognosis results measured in concordance index are indistinguishable across models while are highly variable across cancers. These findings shedding some light into the relationships between patient characteristics and survival learnability on a pan-cancer level.

BERMUDA: a novel deep transfer learning method for single-cell RNA sequencing batch correction reveals hidden high-resolution cellular subtypes.

To fully utilize the power of single-cell RNA sequencing (scRNA-seq) technologies for identifying cell lineages and bona fide transcriptional signals, it is necessary to combine data from multiple experiments. We present BERMUDA (Batch Effect ReMoval Using Deep Autoencoders), a novel transfer-learning-based method for batch effect correction in scRNA-seq data. BERMUDA effectively combines different batches of scRNA-seq data with vastly different cell population compositions and amplifies biological signals by transferring information among batches. We demonstrate that BERMUDA outperforms existing methods for removing batch effects and distinguishing cell types in multiple simulated and real scRNA-seq datasets.

Gene Co-Expression Networks Restructured Gene Fusion in Rhabdomyosarcoma Cancers.

Rhabdomyosarcoma is subclassified by the presence or absence of a recurrent chromosome translocation that fuses the FOXO1 and PAX3 or PAX7 genes. The fusion protein (FOXO1-PAX3/7) retains both binding domains and becomes a novel and potent transcriptional regulator in rhabdomyosarcoma subtypes. Many studies have characterized and integrated genomic, transcriptomic, and epigenomic differences among rhabdomyosarcoma subtypes that contain the FOXO1-PAX3/7 gene fusion and those that do not; however, few investigations have investigated how gene co-expression networks are altered by FOXO1-PAX3/7. Although transcriptional data offer insight into one level of functional regulation, gene co-expression networks have the potential to identify biological interactions and pathways that underpin oncogenesis and tumorigenicity. Thus, we examined gene co-expression networks for rhabdomyosarcoma that were FOXO1-PAX3 positive, FOXO1-PAX7 positive, or fusion negative. Gene co-expression networks were mined using local maximum Quasi-Clique Merger (lmQCM) and analyzed for co-expression differences among rhabdomyosarcoma subtypes. This analysis observed 41 co-expression modules that were shared between fusion negative and positive samples, of which 17/41 showed significant up- or down-regulation in respect to fusion status. Fusion positive and negative rhabdomyosarcoma showed differing modularity of co-expression networks with fusion negative (n = 109) having significantly more individual modules than fusion positive (n = 53). Subsequent analysis of gene co-expression networks for PAX3 and PAX7 type fusions observed 17/53 were differentially expressed between the two subtypes. Gene list enrichment analysis found that gene ontology terms were poorly matched with biological processes and molecular function for most co-expression modules identified in this study; however, co-expressed modules were frequently localized to cytobands on chromosomes 8 and 11. Overall, we observed substantial restructuring of co-expression networks relative to fusion status and fusion type in rhabdomyosarcoma and identified previously overlooked genes and pathways that may be targeted in this pernicious disease.

SALMON: Survival Analysis Learning With Multi-Omics Neural Networks on Breast Cancer.

Improved cancer prognosis is a central goal for precision health medicine. Though many models can predict differential survival from data, there is a strong need for sophisticated algorithms that can aggregate and filter relevant predictors from increasingly complex data inputs. In turn, these models should provide deeper insight into which types of data are most relevant to improve prognosis. Deep Learning-based neural networks offer a potential solution for both problems because they are highly flexible and account for data complexity in a non-linear fashion. In this study, we implement Deep Learning-based networks to determine how gene expression data predicts Cox regression survival in breast cancer. We accomplish this through an algorithm called SALMON (Survival Analysis Learning with Multi-Omics Neural Networks), which aggregates and simplifies gene expression data and cancer biomarkers to enable prognosis prediction. The results revealed improved performance when more omics data were used in model construction. Rather than use raw gene expression values as model inputs, we innovatively use eigengene modules from the result of gene co-expression network analysis. The corresponding high impact co-expression modules and other omics data are identified by feature selection technique, then examined by conducting enrichment analysis and exploiting biological functions, escalated the interpretation of input feature from gene level to co-expression modules level. Our study shows the feasibility of discovering breast cancer related co-expression modules, sketch a blueprint of future endeavors on Deep Learning-based survival analysis. SALMON source code is available at https://github.com/huangzhii/SALMON/.

Micro-computed tomography of pupal metamorphosis in the solitary bee Megachile rotundata.

Insect metamorphosis involves a complex change in form and function. In this study, we examined the development of the solitary bee, Megachile rotundata, using micro-computed tomography (µCT) and volume analysis. We describe volumetric changes of brain, tracheae, flight muscles, gut, and fat bodies in prepupal, pupal, and adult M. rotundata. We observed that individual organ systems have distinct patterns of developmental progression, which vary in their timing and duration. This has important implications for commercial management of this agriculturally relevant pollinator.

The geometric framework for nutrition reveals interactions between protein and carbohydrate during larval growth in honey bees.

In holometabolous insects, larval nutrition affects adult body size, a life history trait with a profound influence on performance and fitness. Individual nutritional components of larval diets are often complex and may interact with one another, necessitating the use of a geometric framework for elucidating nutritional effects. In the honey bee, Apis mellifera, nurse bees provision food to developing larvae, directly moderating growth rates and caste development. However, the eusocial nature of honey bees makes nutritional studies challenging, because diet components cannot be systematically manipulated in the hive. Using in vitro rearing, we investigated the roles and interactions between carbohydrate and protein content on larval survival, growth, and development in A. mellifera We applied a geometric framework to determine how these two nutritional components interact across nine artificial diets. Honey bees successfully completed larval development under a wide range of protein and carbohydrate contents, with the medium protein (∼5%) diet having the highest survival. Protein and carbohydrate both had significant and non-linear effects on growth rate, with the highest growth rates observed on a medium-protein, low-carbohydrate diet. Diet composition did not have a statistically significant effect on development time. These results confirm previous findings that protein and carbohydrate content affect the growth of A. mellifera larvae. However, this study identified an interaction between carbohydrate and protein content that indicates a low-protein, high-carb diet has a negative effect on larval growth and survival. These results imply that worker recruitment in the hive would decline under low protein conditions, even when nectar abundance or honey stores are sufficient.

Evidence of a hemolymph-born factor that induces onset of maturation in Manduca sexta larvae.

Insect metamorphosis is a complex developmental transition determined and coordinated by hormonal signaling that begins at a critical weight late in the larval phase of life. Even though this hormonal signaling is well understood in insects, the internal factors that are assessed at the critical weight and that drive commitment to metamorphosis have remained unresolved in most species. The critical weight may represent either an autonomous decision by the neuroendocrine system without input from other developing larval tissues, or an assessment of developmental thresholds occurring throughout the body that are then integrated by the neuroendocrine tissues. The latter hypothesis predicts that there could be one or more developmental threshold signals that originate from developing tissues and ultimately induce the onset of metamorphosis. However, there is no evidence for such a signal in the organisms for which the critical weight is well described. Here we test for the evidence of this factor in Manduca sexta (Lepidoptera: Sphingidae) by transferring hemolymph from individuals that are either post- or pre-critical weight into pre-critical weight 5(th) instar larvae. We found that hemolymph from a post-critical weight donor induces a shortening of development time, though the mass at pupation is unaffected. This suggests that metamorphic commitment occurring at the critical weight is at least partially coordinated by signaling from developing tissues via a hemolymph-borne signaling factor.

Predicting performance and plasticity in the development of respiratory structures and metabolic systems.

The scaling laws governing metabolism suggest that we can predict metabolic rates across taxonomic scales that span large differences in mass. Yet, scaling relationships can vary with development, body region, and environment. Within species, there is variation in metabolic rate that is independent of mass and which may be explained by genetic variation, the environment or their interaction (i.e., metabolic plasticity). Additionally, some structures, such as the insect tracheal respiratory system, change throughout development and in response to the environment to match the changing functional requirements of the organism. We discuss how study of the development of respiratory function meets multiple challenges set forth by the NSF Grand Challenges Workshop. Development of the structure and function of respiratory and metabolic systems (1) is inherently stable and yet can respond dynamically to change, (2) is plastic and exhibits sensitivity to environments, and (3) can be examined across multiple scales in time and space. Predicting respiratory performance and plasticity requires quantitative models that integrate information across scales of function from the expression of metabolic genes and mitochondrial biogenesis to the building of respiratory structures. We present insect models where data are available on the development of the tracheal respiratory system and of metabolic physiology and suggest what is needed to develop predictive models. Incorporating quantitative genetic data will enable mapping of genetic and genetic-by-environment variation onto phenotypes, which is necessary to understand the evolution of respiratory and metabolic systems and their ability to enable respiratory homeostasis as organisms walk the tightrope between stability and change.