Decoding phenotypic screening: A comparative analysis of image representations.

Biomedical imaging techniques such as high content screening (HCS) are valuable for drug discovery, but high costs limit their use to pharmaceutical companies. To address this issue, The JUMP-CP consortium released a massive open image dataset of chemical and genetic perturbations, providing a valuable resource for deep learning research. In this work, we aim to utilize the JUMP-CP dataset to develop a universal representation model for HCS data, mainly data generated using U2OS cells and CellPainting protocol, using supervised and self-supervised learning approaches. We propose an evaluation protocol that assesses their performance on mode of action and property prediction tasks using a popular phenotypic screening dataset. Results show that the self-supervised approach that uses data from multiple consortium partners provides representation that is more robust to batch effects whilst simultaneously achieving performance on par with standard approaches. Together with other conclusions, it provides recommendations on the training strategy of a representation model for HCS images.

Leveraging Cell Painting Images to Expand the Applicability Domain and Actively Improve Deep Learning Quantitative Structure-Activity Relationship Models.

The search for chemical hit material is a lengthy and increasingly expensive drug discovery process. To improve it, ligand-based quantitative structure-activity relationship models have been broadly applied to optimize primary and secondary compound properties. Although these models can be deployed as early as the stage of molecule design, they have a limited applicability domain─if the structures of interest differ substantially from the chemical space on which the model was trained, a reliable prediction will not be possible. Image-informed ligand-based models partly solve this shortcoming by focusing on the phenotype of a cell caused by small molecules, rather than on their structure. While this enables chemical diversity expansion, it limits the application to compounds physically available and imaged. Here, we employ an active learning approach to capitalize on both of these methods' strengths and boost the model performance of a mitochondrial toxicity assay (Glu/Gal). Specifically, we used a phenotypic Cell Painting screen to build a chemistry-independent model and adopted the results as the main factor in selecting compounds for experimental testing. With the additional Glu/Gal annotation for selected compounds we were able to dramatically improve the chemistry-informed ligand-based model with respect to the increased recognition of compounds from a 10% broader chemical space.

ViLoN-a multi-layer network approach to data integration demonstrated for patient stratification.

With more and more data being collected, modern network representations exploit the complementary nature of different data sources as well as similarities across patients. We here introduce the Variation of information fused Layers of Networks algorithm (ViLoN), a novel network-based approach for the integration of multiple molecular profiles. As a key innovation, it directly incorporates prior functional knowledge (KEGG, GO). In the constructed network of patients, patients are represented by networks of pathways, comprising genes that are linked by common functions and joint regulation in the disease. Patient stratification remains a key challenge both in the clinic and for research on disease mechanisms and treatments. We thus validated ViLoN for patient stratification on multiple data type combinations (gene expression, methylation, copy number), showing substantial improvements and consistently competitive performance for all. Notably, the incorporation of prior functional knowledge was critical for good results in the smaller cohorts (rectum adenocarcinoma: 90, esophageal carcinoma: 180), where alternative methods failed.

Molecular Mechanisms of Fetal Tendon Regeneration Versus Adult Fibrous Repair.

Tendinopathies are painful, disabling conditions that afflict 25% of the adult human population. Filling an unmet need for realistic large-animal models, we here present an ovine model of tendon injury for the comparative study of adult scarring repair and fetal regeneration. Complete regeneration of the fetal tendon within 28 days is demonstrated, while adult tendon defects remained macroscopically and histologically evident five months post-injury. In addition to a comprehensive histological assessment, proteome analyses of secretomes were performed. Confirming histological data, a specific and pronounced inflammation accompanied by activation of neutrophils in adult tendon defects was observed, corroborated by the significant up-regulation of pro-inflammatory factors, neutrophil attracting chemokines, the release of potentially tissue-damaging antimicrobial and extracellular matrix-degrading enzymes, and a response to oxidative stress. In contrast, secreted proteins of injured fetal tendons included proteins initiating the resolution of inflammation or promoting functional extracellular matrix production. These results demonstrate the power and relevance of our novel ovine fetal tendon regeneration model, which thus promises to accelerate research in the field. First insights from the model already support our molecular understanding of successful fetal tendon healing processes and may guide improved therapeutic strategies.

The in vivo genetic program of murine primordial lung epithelial progenitors.

Multipotent Nkx2-1-positive lung epithelial primordial progenitors of the foregut endoderm are thought to be the developmental precursors to all adult lung epithelial lineages. However, little is known about the global transcriptomic programs or gene networks that regulate these gateway progenitors in vivo. Here we use bulk RNA-sequencing to describe the unique genetic program of in vivo murine lung primordial progenitors and computationally identify signaling pathways, such as Wnt and Tgf-ß superfamily pathways, that are involved in their cell-fate determination from pre-specified embryonic foregut. We integrate this information in computational models to generate in vitro engineered lung primordial progenitors from mouse pluripotent stem cells, improving the fidelity of the resulting cells through unbiased, easy-to-interpret similarity scores and modulation of cell culture conditions, including substratum elastic modulus and extracellular matrix composition. The methodology proposed here can have wide applicability to the in vitro derivation of bona fide tissue progenitors of all germ layers.

Sex-Specific Human Cardiomyocyte Gene Regulation in Left Ventricular Pressure Overload.

OBJECTIVE: To assess gene expression in cardiomyocytes isolated from patients with aortic stenosis, hypothesizing that maladaptive remodeling and inflammation-related genes are higher in male vs female patients. PATIENTS AND METHODS: In this study, 34 patients with aortic stenosis undergoing aortic valve replacement from March 20, 2016, through May 24, 2017, at the German Heart Centre in Berlin, Germany, were included. Isolated cardiomyocytes from interventricular septum samples were used for gene expression analysis. Clinical and echocardiographic data were collected preoperatively. RESULTS: Age, body mass index, systolic and diastolic blood pressure, comorbidities, and medication were similar between the 17 male and 17 female patients. The mean ± SD left ventricular end-diastolic diameter (52±9 vs 45±4 mm; P=.007) and posterior wall thickness (14.2±2.5 vs 12.1±1.6 mm; P=.03) were higher in male vs female patients, while ejection fraction was lower in male patients (49%±14% vs 59%±5%; P=.01). Focusing on structural genes involved in the development of cardiac hypertrophy and remodeling, we found that most were expressed higher in male vs female patients. Our modeling analysis revealed that 2 inflammation-related genes, CCN2 and NFKB1, were negatively related to ejection fraction, with this effect being male specific (P=.03 and P=.02, respectively). CONCLUSION: These findings provide novel insight into cardiomyocyte-specific molecular changes related to sex differences in pressure overload and a significant male-specific association between cardiac function and inflammation-related genes. Considering these sex differences may contribute toward a more accurate design of research and the development of more appropriate therapeutic approaches for both male and female patients.

A novel framework for horizontal and vertical data integration in cancer studies with application to survival time prediction models.

BACKGROUND: Recently high-throughput technologies have been massively used alongside clinical tests to study various types of cancer. Data generated in such large-scale studies are heterogeneous, of different types and formats. With lack of effective integration strategies novel models are necessary for efficient and operative data integration, where both clinical and molecular information can be effectively joined for storage, access and ease of use. Such models, combined with machine learning methods for accurate prediction of survival time in cancer studies, can yield novel insights into disease development and lead to precise personalized therapies. RESULTS: We developed an approach for intelligent data integration of two cancer datasets (breast cancer and neuroblastoma) - provided in the CAMDA 2018 'Cancer Data Integration Challenge', and compared models for prediction of survival time. We developed a novel semantic network-based data integration framework that utilizes NoSQL databases, where we combined clinical and expression profile data, using both raw data records and external knowledge sources. Utilizing the integrated data we introduced Tumor Integrated Clinical Feature (TICF) - a new feature for accurate prediction of patient survival time. Finally, we applied and validated several machine learning models for survival time prediction. CONCLUSION: We developed a framework for semantic integration of clinical and omics data that can borrow information across multiple cancer studies. By linking data with external domain knowledge sources our approach facilitates enrichment of the studied data by discovery of internal relations. The proposed and validated machine learning models for survival time prediction yielded accurate results. REVIEWERS: This article was reviewed by Eran Elhaik, Wenzhong Xiao and Carlos Loucera.

Fetal articular cartilage regeneration versus adult fibrocartilaginous repair: secretome proteomics unravels molecular mechanisms in an ovine model.

Osteoarthritis (OA), a degenerative joint disease characterized by progressive cartilage degeneration, is one of the leading causes of disability worldwide owing to the limited regenerative capacity of adult articular cartilage. Currently, there are no disease-modifying pharmacological or surgical therapies for OA. Fetal mammals, in contrast to adults, are capable of regenerating injured cartilage in the first two trimesters of gestation. A deeper understanding of the properties intrinsic to the response of fetal tissue to injury would allow us to modulate the way in which adult tissue responds to injury. In this study, we employed secretome proteomics to compare fetal and adult protein regulation in response to cartilage injury using an ovine cartilage defect model. The most relevant events comprised proteins associated with the immune response and inflammation, proteins specific for cartilage tissue and cartilage development, and proteins involved in cell growth and proliferation. Alarmins S100A8, S100A9 and S100A12 and coiled-coil domain containing 88A (CCDC88A), which are associated with inflammatory processes, were found to be significantly upregulated following injury in adult, but not in fetal animals. By contrast, cartilage-specific proteins like proteoglycan 4 were upregulated in response to injury only in fetal sheep postinjury. Our results demonstrate the power and relevance of the ovine fetal cartilage regeneration model presented here for the first time. The identification of previously unrecognized modulatory proteins that plausibly affect the healing process holds great promise for potential therapeutic interventions.

Transcriptomic changes in CHO cells after adaptation to suspension growth in protein-free medium analysed by a species-specific microarray.

Chinese Hamster Ovary (CHO) cells are the preferred cell line for production of biopharmaceuticals. These cells are capable to grow without serum supplementation, but drastic changes in their phenotype occur during adaptation to protein-free growth, which typically include the change to a suspension phenotype with reduced growth rate. A possible approach to understand this transformation, with the intention to counteract the reduction in growth by targeted supplementation of protein-free media, is gene expression profiling. The increasing availability of genome-scale data for CHO now facilitates quests for a better understanding of metabolic pathways and gene networks. So far, systematic large-scale expression profiling in CHO cells by microarray was limited due to lack of publicly available array designs and limitations of alternative approaches. Based on the recent release of CHO and Chinese Hamster genome sequences, including an annotated RefSeq genome, we have constructed a publicly available microarray design for effective genome-scale expression profiling. The design employed microarray probes optimized for uniformity, sensitivity, and specificity, with probe properties computed using the latest thermodynamic models. We validated the platform in an analysis of gene expression changes in response to serum-free adaptation. The observed effects on the lipid metabolism as well as on nucleotide synthesis were used to successfully select media supplements that were able to increase growth rate.

Thyroid Progenitors Are Robustly Derived from Embryonic Stem Cells through Transient, Developmental Stage-Specific Overexpression of Nkx2-1.

The clinical importance of anterior foregut endoderm (AFE) derivatives, such as thyrocytes, has led to intense research efforts for their derivation through directed differentiation of pluripotent stem cells (PSCs). Here, we identify transient overexpression of the transcription factor (TF) NKX2-1 as a powerful inductive signal for the robust derivation of thyrocyte-like cells from mouse PSC-derived AFE. This effect is highly developmental stage specific and dependent on FOXA2 expression levels and precise modulation of BMP and FGF signaling. The majority of the resulting cells express thyroid TFs (Nkx2-1, Pax8, Foxe1, Hhex) and thyroid hormone synthesis-related genes (Tg, Tpo, Nis, Iyd) at levels similar to adult mouse thyroid and give rise to functional follicle-like epithelial structures in Matrigel culture. Our findings demonstrate that NKX2-1 overexpression converts AFE to thyroid epithelium in a developmental time-sensitive manner and suggest a general methodology for manipulation of cell-fate decisions of developmental intermediates.

Polarization-Sensitive Optical Coherence Tomography and Conventional Retinal Imaging Strategies in Assessing Foveal Integrity in Geographic Atrophy.

PURPOSE: To compare current imaging methods with respect to their ability to detect the condition of the fovea in patients with geographic atrophy (GA). METHODS: The retinas of 176 eyes with GA were imaged using two spectral-domain optical coherence tomography (SD-OCT) systems, Cirrus HD-OCT and Spectralis HRA+OCT, and fundus autofluorescence (FAF) and infrared imaging (IR) was used in the scanning laser ophthalmoscope (SLO) mode. Polarization-sensitive OCT (PS-OCT), which selectively visualizes the RPE in addition to SD-OCT features, was used to image 95 eyes. Geographic atrophy lesions were categorized as fovea spared, involved, or not quantifiable (grades 0, 1, and 2). Morphologic gradings were subsequently correlated with best-corrected visual acuity (BCVA) measurements to independently identify the corresponding functional condition of the fovea. Cohen's κ statistics with a bootstrap method was applied to compare retinal imaging methods. RESULTS: In PS-OCT, 84% of eyes with BCVA greater than or equal to 20/40 were detected, whereas in conventional retinal imaging the rate ranged from 27% in FAF to 45% in the SD-OCT segment. Cohen's κ statistics revealed significant differences between the gradings of PS-OCT and conventional imaging with κ = 0.488 and a global Hotelling's T2 statistic of 17.9 with a P value of P = 0.003. Statistical tests revealed no statistically significant differences between the conventional retinal imaging modalities. CONCLUSIONS: Polarization-sensitive OCT can better allow correct grading of the fovea in relation to BCVA and identify foveal sparing than other imaging modalities. The differences in imaging precision should be considered in diagnostic and therapeutic evaluations.

Experiences with workflows for automating data-intensive bioinformatics.

High-throughput technologies, such as next-generation sequencing, have turned molecular biology into a data-intensive discipline, requiring bioinformaticians to use high-performance computing resources and carry out data management and analysis tasks on large scale. Workflow systems can be useful to simplify construction of analysis pipelines that automate tasks, support reproducibility and provide measures for fault-tolerance. However, workflow systems can incur significant development and administration overhead so bioinformatics pipelines are often still built without them. We present the experiences with workflows and workflow systems within the bioinformatics community participating in a series of hackathons and workshops of the EU COST action SeqAhead. The organizations are working on similar problems, but we have addressed them with different strategies and solutions. This fragmentation of efforts is inefficient and leads to redundant and incompatible solutions. Based on our experiences we define a set of recommendations for future systems to enable efficient yet simple bioinformatics workflow construction and execution.