Jiande Wu

The Covid-19 pandemic has highlighted the urgent need for accurate prediction of pandemic trends. We propose a deep learning model for predicting Covid-19 cases and deaths at the county level through transformer neural networks with multi-source data fusion, incorporating historical case data, death data, and social media sentiment analysis to capture both temporal (historical trends) and spatial (geographical relationships) dynamics within time series data. Additionally, we develop multi-level and multi-scale attention mechanisms for adaptive time-frequency analysis. Across three Omicron variant waves (December 2021 through February 2023), the model demonstrated strong performance in predicting county-level Covid-19 cases and deaths, with median county agreement accuracy ranging from 74.0% to 82.6% for one-week case forecasts and 68.7% to 79.6% for 5-week case forecasts. Median county agreement accuracy for deaths ranged from 83.2% to 86.3% for one-week forecasts and 84.3% to 87.2% for five-week forecasts. Incorporating social media data yielded mild to moderate improvement in forecasting accuracy. Overall, the proposed model yielded substantial improvements compared to a baseline persistence model utilizing the last observation carried forward. By integrating real-time data and capturing complex pandemic dynamics, this approach surpasses traditional methods. Its high accuracy and generalizability make it a valuable tool for enhancing public health preparedness and response strategies in future outbreaks.

A critical unmet medical need in prostate cancer (PCa) clinical management centers around distinguishing indolent from aggressive tumors. Traditionally, Gleason grading has been utilized for this purpose. However, tumor classification using Gleason Grade 7 is often ambiguous, as the clinical behavior of these tumors follows a variable clinical course. This study aimed to investigate the application of machine learning techniques (ML) to classify patients into indolent and aggressive PCas. We used gene expression data from The Cancer Genome Atlas and compared gene expression levels between indolent and aggressive tumors to identify features for developing and validating a range of ML and stacking algorithms. ML algorithms accurately distinguished indolent from aggressive PCas. With the accuracy of 96%, the stacking model was superior to individual ML algorithms when all samples with primary Gleason Grades 6 to 10 were used. Excluding samples with Gleason Grade 7 improved accuracy to 97%. This study shows that ML algorithms and stacking models are powerful approaches for the accurate classification of indolent versus aggressive PCas. Future implementation of this methodology may significantly impact clinical decision making and patient outcomes in the clinical management of prostate cancer.

Accurately distinguishing between indolent and aggressive tumors is a crucial unmet need in the clinical management of prostate cancer (PCa). The traditional Gleason grading system has been utilized for this purpose; however, there is often ambiguity in classifying tumors with a Gleason grade of 7. Clinicians commonly resort to using secondary Gleason grades, such as 3+4 or 4+3, to classify these tumors as indolent or aggressive, respectively. Unfortunately, such classifications are prone to misinterpretation, leading to erroneous diagnoses and prognoses. To address this challenge, we investigated the application of Machine Learning (ML) techniques to classify PCa patients based on gene expression data sourced from The Cancer Genome Atlas. By comparing gene expression levels between indolent and aggressive tumors, we sought to identify distinctive features for developing and validating a range of ML algorithms and stacking techniques. The stacking based model achieved an impressive accuracy of 96% for all samples encompassing primary Gleason grades 6 to 10. Notably, when excluding Gleason grade 7 from the analysis, the accuracy further improved to 97%. This study underscores the effectiveness of the stacked ML algorithm for accurately classifying indolent versus aggressive PCa. Leveraging gene expression data and employing a combination of classifiers, this approach offers a powerful solution to address the unmet need in robustly distinguishing between different types of PCa tumors. Future implementation of this methodology may significantly impact clinical decision-making and patient outcomes in the management of prostate cancer.

Background: A critical unmet medical need in clinical management of colorectal cancer (CRC) pivots around lack of noninvasive and or minimally invasive techniques for early diagnosis and prognostic prediction of clinical outcomes. Because DNA methylation can capture the regulatory landscape of tumors and can be measured in body fluids, it provides unparalleled opportunities for the discovery of early diagnostic and prognostics markers predictive of clinical outcomes. Here we investigated use of DNA methylation for the discovery of potential clinically actionable diagnostic and prognostic markers for predicting survival in CRC. Methods: We analyzed DNA methylation patterns between tumor and control samples to discover signatures of CpG sites and genes associated with CRC and predictive of survival. We conducted functional analysis to identify molecular networks and signaling pathways driving clinical outcomes. Results: We discovered a signature of aberrantly methylated genes associated with CRC and a signature of thirteen (13) CpG sites predictive of survival. We discovered molecular networks and signaling pathways enriched for CpG sites likely to drive clinical outcomes. Conclusions: The investigation revealed that CpG sites can predict survival in CRC and that DNA methylation can capture the regulatory state of tumors through aberrantly methylated molecular networks and signaling pathways.

Background: the development and progression of triple-negative breast cancer (TNBC) is driven by somatic driver mutations and the tumor-immune microenvironment. To date, data on somatic mutations has not been leveraged and integrated with information on the immune microenvironment to elucidate the possible oncogenic interactions and their potential effects on clinical outcomes. Here, we investigated possible oncogenic interactions between somatic mutations and the tumor-immune microenvironment, and their correlation with patient survival in TNBC. Methods: We performed analysis combining data on 7,875 somatic mutated genes with information on 1,751 immune-modulated genes, using gene-expression data as the intermediate phenotype, and correlated the resulting information with survival. We conducted functional analysis to identify immune-modulated molecular networks and signaling pathways enriched for somatic mutations likely to drive clinical outcomes. Results: We discovered differences in somatic mutation profiles between patients who died and those who survived, and a signature of somatic mutated immune-modulated genes transcriptionally associated with TNBC, predictive of survival. In addition, we discovered immune-modulated molecular networks and signaling pathways enriched for somatic mutations. Conclusions: The investigation revealed possible oncogenic interactions between somatic mutations and the tumor-immune microenvironment in TNBC, likely to affect clinical outcomes.

Background: Lung cancer patients have the worst outcomes when affected by coronavirus disease 2019 (COVID-19). The molecular mechanisms underlying the association between lung cancer and COVID-19 remain unknown. The objective of this investigation was to determine whether there is crosstalk in molecular perturbation between COVID-19 and lung cancer, and to identify a molecular signature, molecular networks and signaling pathways shared by the two diseases. Methods: We analyzed publicly available gene expression data from 52 severely affected COVID-19 human lung samples, 594 lung tumor samples and 54 normal disease-free lung samples. We performed network and pathways analysis to identify molecular networks and signaling pathways shared by the two diseases. Results: The investigation revealed a signature of genes associated with both diseases and signatures of genes uniquely associated with each disease, confirming crosstalk in molecular perturbation between COVID-19 and lung cancer. In addition, the analysis revealed molecular networks and signaling pathways associated with both diseases. Conclusions: The investigation revealed crosstalk in molecular perturbation between COVID-19 and lung cancer, and molecular networks and signaling pathways associated with the two diseases. Further research on a population impacted by both diseases is recommended to elucidate molecular drivers of the association between the two diseases.

Background: Breast cancer is a heterogeneous disease defined by molecular types and subtypes. Advances in genomic research have enabled use of precision medicine in clinical management of breast cancer. A critical unmet medical need is distinguishing triple negative breast cancer, the most aggressive and lethal form of breast cancer, from non-triple negative breast cancer. Here we propose use of a machine learning (ML) approach for classification of triple negative breast cancer and non-triple negative breast cancer patients using gene expression data. Methods: We performed analysis of RNA-Sequence data from 110 triple negative and 992 non-triple negative breast cancer tumor samples from The Cancer Genome Atlas to select the features (genes) used in the development and validation of the classification models. We evaluated four different classification models including Support Vector Machines, K-nearest neighbor, Naïve Bayes and Decision tree using features selected at different threshold levels to train the models for classifying the two types of breast cancer. For performance evaluation and validation, the proposed methods were applied to independent gene expression datasets. Results: Among the four ML algorithms evaluated, the Support Vector Machine algorithm was able to classify breast cancer more accurately into triple negative and non-triple negative breast cancer and had less misclassification errors than the other three algorithms evaluated. Conclusions: The prediction results show that ML algorithms are efficient and can be used for classification of breast cancer into triple negative and non-triple negative breast cancer types.

Background: The recent surge of next generation sequencing of breast cancer genomes has enabled development of comprehensive catalogues of somatic mutations and expanded the molecular classification of subtypes of breast cancer. However, somatic mutations and gene expression data have not been leveraged and integrated with epigenomic data to unravel the genomic-epigenomic interaction landscape of triple negative breast cancer (TNBC) and non-triple negative breast cancer (non-TNBC). Methods: We performed integrative data analysis combining somatic mutation, epigenomic and gene expression data from The Cancer Genome Atlas (TCGA) to unravel the possible oncogenic interactions between genomic and epigenomic variation in TNBC and non-TNBC. We hypothesized that within breast cancers, there are differences in somatic mutation, DNA methylation and gene expression signatures between TNBC and non-TNBC. We further hypothesized that genomic and epigenomic alterations affect gene regulatory networks and signaling pathways driving the two types of breast cancer. Results: The investigation revealed somatic mutated, epigenomic and gene expression signatures unique to TNBC and non-TNBC and signatures distinguishing the two types of breast cancer. In addition, the investigation revealed molecular networks and signaling pathways enriched for somatic mutations and epigenomic changes unique to each type of breast cancer. The most significant pathways for TNBC were: retinal biosynthesis, BAG2, LXR/RXR, EIF2 and P2Y purigenic receptor signaling pathways. The most significant pathways for non-TNBC were: UVB-induced MAPK, PCP, Apelin endothelial, Endoplasmatic reticulum stress and mechanisms of viral exit from host signaling Pathways. Conclusion: The investigation revealed integrated genomic, epigenomic and gene expression signatures and signing pathways unique to TNBC and non-TNBC, and a gene signature distinguishing the two types of breast cancer. The study demonstrates that integrative analysis of multi-omics data is a powerful approach for unravelling the genomic-epigenomic interaction landscape in TNBC and non-TNBC.

Triple-negative breast cancer (TNBC) is the most aggressive form of breast cancer. Emerging evidenced suggests that both genetics and epigenetic factors play a role in the pathogenesis of TNBC. However, oncogenic interactions and cooperation between genomic and epigenomic variation have not been characterized. The objective of this study was to deconvolute the genomic and epigenomic interaction landscape in TNBC using an integrative genomics approach, which integrates information on germline, somatic, epigenomic and gene expression variation. We hypothesized that TNBC originates from a complex interplay between genomic (both germline and somatic variation) and epigenomic variation. We further hypothesized that these complex arrays of interacting genomic and epigenomic factors affect entire molecular networks and signaling pathways which, in turn, drive TNBC. We addressed these hypotheses using germline variation from genome-wide association studies and somatic, epigenomic and gene expression variation from The Cancer Genome Atlas (TCGA). The investigation revealed signatures of functionally related genes containing germline, somatic and epigenetic variations. DNA methylation had an effect on gene expression. Network and pathway analysis revealed molecule networks and signaling pathways enriched for germline, somatic and epigenomic variation, among them: Role of BRCA1 in DNA Damage Response, Hereditary Breast Cancer Signaling, Molecular Mechanisms of Cancer, Estrogen-Dependent Breast Cancer, p53, MYC Mediated Apoptosis, and PTEN Signaling pathways. The investigation revealed that integrative genomics is a powerful approach for deconvoluting the genomic-epigenomic interaction landscape in TNBC. Further studies are needed to understand the biological mechanisms underlying oncogenic interactions between genomic and epigenomic factors in TNBC.