Digital profiling of cancer transcriptomes from histology images with grouped vision attention.

Cancer is a heterogeneous disease that demands precise molecular profiling for better understanding and management. Recently, deep learning has demonstrated potentials for cost-efficient prediction of molecular alterations from histology images. While transformer-based deep learning architectures have enabled significant progress in non-medical domains, their application to histology images remains limited due to small dataset sizes coupled with the explosion of trainable parameters. Here, we develop SEQUOIA, a transformer model to predict cancer transcriptomes from whole-slide histology images. To enable the full potential of transformers, we first pre-train the model using data from 1,802 normal tissues. Then, we fine-tune and evaluate the model in 4,331 tumor samples across nine cancer types. The prediction performance is assessed at individual gene levels and pathway levels through Pearson correlation analysis and root mean square error. The generalization capacity is validated across two independent cohorts comprising 1,305 tumors. In predicting the expression levels of 25,749 genes, the highest performance is observed in cancers from breast, kidney and lung, where SEQUOIA accurately predicts the expression of 11,069, 10,086 and 8,759 genes, respectively. The accurately predicted genes are associated with the regulation of inflammatory response, cell cycles and metabolisms. While the model is trained at the tissue level, we showcase its potential in predicting spatial gene expression patterns using spatial transcriptomics datasets. Leveraging the prediction performance, we develop a digital gene expression signature that predicts the risk of recurrence in breast cancer. SEQUOIA deciphers clinically relevant gene expression patterns from histology images, opening avenues for improved cancer management and personalized therapies.

Figure Correction: Using Social Media to Help Understand Patient-Reported Health Outcomes of Post-COVID-19 Condition: Natural Language Processing Approach.

[This corrects the article DOI: 10.2196/45767.].

Graph-based multi-modality integration for prediction of cancer subtype and severity.

Personalised cancer screening before therapy paves the way toward improving diagnostic accuracy and treatment outcomes. Most approaches are limited to a single data type and do not consider interactions between features, leaving aside the complementary insights that multimodality and systems biology can provide. In this project, we demonstrate the use of graph theory for data integration via individual networks where nodes and edges are individual-specific. We showcase the consequences of early, intermediate, and late graph-based fusion of RNA-Seq data and histopathology whole-slide images for predicting cancer subtypes and severity. The methodology developed is as follows: (1) we create individual networks; (2) we compute the similarity between individuals from these graphs; (3) we train our model on the similarity matrices; (4) we evaluate the performance using the macro F1 score. Pros and cons of elements of the pipeline are evaluated on publicly available real-life datasets. We find that graph-based methods can increase performance over methods that do not study interactions. Additionally, merging multiple data sources often improves classification compared to models based on single data, especially through intermediate fusion. The proposed workflow can easily be adapted to other disease contexts to accelerate and enhance personalized healthcare.

Using Social Media to Help Understand Patient-Reported Health Outcomes of Post-COVID-19 Condition: Natural Language Processing Approach.

BACKGROUND: While scientific knowledge of post-COVID-19 condition (PCC) is growing, there remains significant uncertainty in the definition of the disease, its expected clinical course, and its impact on daily functioning. Social media platforms can generate valuable insights into patient-reported health outcomes as the content is produced at high resolution by patients and caregivers, representing experiences that may be unavailable to most clinicians. OBJECTIVE: In this study, we aimed to determine the validity and effectiveness of advanced natural language processing approaches built to derive insight into PCC-related patient-reported health outcomes from social media platforms Twitter and Reddit. We extracted PCC-related terms, including symptoms and conditions, and measured their occurrence frequency. We compared the outputs with human annotations and clinical outcomes and tracked symptom and condition term occurrences over time and locations to explore the pipeline's potential as a surveillance tool. METHODS: We used bidirectional encoder representations from transformers (BERT) models to extract and normalize PCC symptom and condition terms from English posts on Twitter and Reddit. We compared 2 named entity recognition models and implemented a 2-step normalization task to map extracted terms to unique concepts in standardized terminology. The normalization steps were done using a semantic search approach with BERT biencoders. We evaluated the effectiveness of BERT models in extracting the terms using a human-annotated corpus and a proximity-based score. We also compared the validity and reliability of the extracted and normalized terms to a web-based survey with more than 3000 participants from several countries. RESULTS: UmlsBERT-Clinical had the highest accuracy in predicting entities closest to those extracted by human annotators. Based on our findings, the top 3 most commonly occurring groups of PCC symptom and condition terms were systemic (such as fatigue), neuropsychiatric (such as anxiety and brain fog), and respiratory (such as shortness of breath). In addition, we also found novel symptom and condition terms that had not been categorized in previous studies, such as infection and pain. Regarding the co-occurring symptoms, the pair of fatigue and headaches was among the most co-occurring term pairs across both platforms. Based on the temporal analysis, the neuropsychiatric terms were the most prevalent, followed by the systemic category, on both social media platforms. Our spatial analysis concluded that 42% (10,938/26,247) of the analyzed terms included location information, with the majority coming from the United States, United Kingdom, and Canada. CONCLUSIONS: The outcome of our social media-derived pipeline is comparable with the results of peer-reviewed articles relevant to PCC symptoms. Overall, this study provides unique insights into patient-reported health outcomes of PCC and valuable information about the patient's journey that can help health care providers anticipate future needs. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): RR2-10.1101/2022.12.14.22283419.

Transcriptional profiling of recall responses to Francisella live vaccine strain.

Global gene expression profile changes were monitored in human peripheral blood mononuclear cells (PBMCs) after challenge with the live vaccine strain (LVS) of Francisella tularensis. Because these PBMCs were from individuals previously immunized with LVS, stimulating these cells with LVS should activate memory responses. The Ingenuity Pathway Analysis tool identified pathways, functions, and networks associated with this in vitro recall response, including novel pathways triggered by the memory response. Dendritic cell (DC) maturation was the most significant among the more than 25 relevant pathways discovered. Interleukin 15, granulocyte-macrophage colony-stimulating factor, and triggering receptor expressed on myeloid cells 1 signaling pathways were also significant. Pathway analysis indicated that Class 1 antigen presentation may not be optimal with LVS vaccination. The top three biological functions were antigen presentation, cell-mediated and humoral immune responses. Network analysis revealed that the top network associated with these functions had IFNγ and TNFα in central interactive positions. Our results suggest that DC maturation is a key factor in the recall responses and that more effective antigen processing and presentation is needed for cytotoxic T lymphocyte responses. Taken together, these considerations are critical for future tularemia vaccine development studies.

Actein activates stress- and statin-associated responses and is bioavailable in Sprague-Dawley rats.

The purpose of this study was to assess in rats the pharmacological parameters and effects on gene expression in the liver of the triterpene glycoside actein. Actein, an active component from the herb black cohosh, has been shown to inhibit the proliferation of human breast cancer cells. To conduct our assessment, we determined the molecular effects of actein on livers from Sprague-Dawley rats treated with actein at 35.7 mg/kg for 6 and 24 h. Chemogenomic analyses indicated that actein elicited stress and statin-associated responses in the liver; actein altered expression of cholesterol and fatty acid biosynthetic genes, p53 pathway genes, CCND1 and ID3. Real-time RT-PCR validated that actein induced three time-dependent patterns of gene expression in the liver: (i) a decrease followed by a significant increase of HMGCS1, HMGCR, HSD17B7, NQO1, S100A9; (ii) a progressive increase of BZRP and CYP7A1 and (iii) a significant increase followed by a decrease of CCND1 and ID3. Consistent with actein's statin- and stress-associated responses, actein reduced free fatty acid and cholesterol content in the liver by 0.6-fold at 24 h and inhibited the growth of human HepG2 liver cancer cells. To determine the bioavailability of actein, we collected serum samples for pharmacokinetic analysis at various times up to 24 h. The serum level of actein peaked at 2.4 microg/mL at 6 h. Actein's ability to alter pathways involved in lipid disorders and carcinogenesis may make it a new agent for preventing and treating these major disorders.

Pathway analysis tools and toxicogenomics reference databases for risk assessment.

The pharmaceutical industry has begun to leverage a range of new technologies (proteomics, pharmacogenomics, metabolomics and molecular toxicology [e.g., toxicogenomics]) and analysis tools that are becoming increasingly integrated in the area of drug discovery and development. The approach of analyzing the vast amount of toxicogenomics data generated using molecular pathway and networks analysis tools in combination with analysis of reference data will be the main focus of this review. We will demonstrate how this combined approach can increase the understanding of the molecular mechanisms that lead to chemical-induced toxicity and application of this knowledge to compound risk assessment. We will provide an example of the insights achieved through a molecular toxicology analysis based on the well-known hepatotoxicant lipopolysaccharide to illustrate the utility of these new tools in the analysis of complex data sets, both in vivo and in vitro. The ultimate objective is a better lead selection process that improves the chances for success across the different stages of drug discovery and development.

Development of a large-scale chemogenomics database to improve drug candidate selection and to understand mechanisms of chemical toxicity and action.

Successful drug discovery requires accurate decision making in order to advance the best candidates from initial lead identification to final approval. Chemogenomics, the use of genomic tools in pharmacology and toxicology, offers a promising enhancement to traditional methods of target identification/validation, lead identification, efficacy evaluation, and toxicity assessment. To realize the value of chemogenomics information, a contextual database is needed to relate the physiological outcomes induced by diverse compounds to the gene expression patterns measured in the same animals. Massively parallel gene expression characterization coupled with traditional assessments of drug candidates provides additional, important mechanistic information, and therefore a means to increase the accuracy of critical decisions. A large-scale chemogenomics database developed from in vivo treated rats provides the context and supporting data to enhance and accelerate accurate interpretation of mechanisms of toxicity and pharmacology of chemicals and drugs. To date, approximately 600 different compounds, including more than 400 FDA approved drugs, 60 drugs approved in Europe and Japan, 25 withdrawn drugs, and 100 toxicants, have been profiled in up to 7 different tissues of rats (representing over 3200 different drug-dose-time-tissue combinations). Accomplishing this task required evaluating and improving a number of in vivo and microarray protocols, including over 80 rigorous quality control steps. The utility of pairing clinical pathology assessments with gene expression data is illustrated using three anti-neoplastic drugs: carmustine, methotrexate, and thioguanine, which had similar effects on the blood compartment, but diverse effects on hepatotoxicity. We will demonstrate that gene expression events monitored in the liver can be used to predict pathological events occurring in that tissue as well as in hematopoietic tissues.

Notch-induced E2A ubiquitination and degradation are controlled by MAP kinase activities.

Notch signals are important for lymphocyte development but downstream events that follow Notch signaling are not well understood. Here, we report that signaling through Notch modulates the turnover of E2A proteins including E12 and E47, which are basic helix-loop-helix proteins crucial for B and T lymphocyte development. Notch-induced degradation requires phosphorylation of E47 by p42/p44 MAP kinases. Expression of the intracellular domain of Notch1 (N1-IC) enhances the association of E47 with the SCF(Skp2) E3 ubiquitin ligase and ubiquitination of E47, followed by proteasome-mediated degradation. Furthermore, N1-IC induces E2A degradation in B and T cells in the presence of activated MAP kinases. Activation of endogenous Notch receptors by treatment of splenocytes with anti-IgM or anti-CD3 plus anti-CD28 also leads to E2A degradation, which is blocked by the inhibitors of Notch activation or proteasome function. Notch-induced E2A degradation depends on the function of its downstream effector, RBP-Jkappa, probably to activate target genes involved in the ubiquitination of E2A proteins. Thus we propose that Notch regulates lymphocyte differentiation by controlling E2A protein turnover.