Multi-modal molecular determinants of clinically relevant osteoporosis subtypes.

Due to a rapidly aging global population, osteoporosis and the associated risk of bone fractures have become a wide-spread public health problem. However, osteoporosis is very heterogeneous, and the existing standard diagnostic measure is not sufficient to accurately identify all patients at risk of osteoporotic fractures and to guide therapy. Here, we constructed the first prospective multi-omics atlas of the largest osteoporosis cohort to date (longitudinal data from 366 participants at three time points), and also implemented an explainable data-intensive analysis framework (DLSF: Deep Latent Space Fusion) for an omnigenic model based on a multi-modal approach that can capture the multi-modal molecular signatures (M3S) as explicit functional representations of hidden genotypes. Accordingly, through DLSF, we identified two subtypes of the osteoporosis population in Chinese individuals with corresponding molecular phenotypes, i.e., clinical intervention relevant subtypes (CISs), in which bone mineral density benefits response to calcium supplements in 2-year follow-up samples. Many snpGenes associated with these molecular phenotypes reveal diverse candidate biological mechanisms underlying osteoporosis, with xQTL preferences of osteoporosis and its subtypes indicating an omnigenic effect on different biological domains. Finally, these two subtypes were found to have different relevance to prior fracture and different fracture risk according to 4-year follow-up data. Thus, in clinical application, M3S could help us further develop improved diagnostic and treatment strategies for osteoporosis and identify a new composite index for fracture prediction, which were remarkably validated in an independent cohort (166 participants).

Identifying therapeutic biomarkers of zoledronic acid by metabolomics.

Zoledronic acid (ZOL) is a potent antiresorptive agent that increases bone mineral density (BMD) and reduces fracture risk in postmenopausal osteoporosis (PMOP). The anti-osteoporotic effect of ZOL is determined by annual BMD measurement. In most cases, bone turnover markers function as early indicators of therapeutic response, but they fail to reflect long-term effects. We used untargeted metabolomics to characterize time-dependent metabolic shifts in response to ZOL and to screen potential therapeutic markers. In addition, bone marrow RNA-seq was performed to support plasma metabolic profiling. Sixty rats were assigned to sham-operated group (SHAM, n = 21) and ovariectomy group (OVX, n = 39) and received sham operation or bilateral ovariectomy, respectively. After modeling and verification, rats in the OVX group were further divided into normal saline group (NS, n = 15) and ZOL group (ZA, n = 18). Three doses of 100 µg/kg ZOL were administrated to the ZA group every 2 weeks to simulate 3-year ZOL therapy in PMOP. An equal volume of saline was administered to the SHAM and NS groups. Plasma samples were collected at five time points for metabolic profiling. At the end of the study, selected rats were euthanatized for bone marrow RNA-seq. A total number of 163 compound were identified as differential metabolites between the ZA and NS groups, including mevalonate, a critical molecule in target pathway of ZOL. In addition, prolyl hydroxyproline (PHP), leucyl hydroxyproline (LHP), 4-vinylphenol sulfate (4-VPS) were identified as differential metabolites throughout the study. Moreover, 4-VPS negatively correlated with increased vertebral BMD after ZOL administration as time-series analysis revealed. Bone marrow RNA-seq showed that the PI3K-AKT signaling pathway was significantly associated with ZOL-mediated changes in expression (adjusted-p = 0.018). In conclusion, mevalonate, PHP, LHP, and 4-VPS are candidate therapeutic markers of ZOL. The pharmacological effect of ZOL likely occurs through inhibition of the PI3K-AKT signaling pathway.

Noninvasive detection and interpretation of gastrointestinal diseases by collaborative serum metabolite and magnetically controlled capsule endoscopy.

Gastrointestinal diseases are complex diseases that occur in the gastrointestinal tract. Common gastrointestinal diseases include chronic gastritis, peptic ulcers, inflammatory bowel disease, and gastrointestinal tumors. These diseases may manifest a long course, difficult treatment, and repeated attacks. Gastroscopy and mucosal biopsy are the gold standard methods for diagnosing gastric and duodenal diseases, but they are invasive procedures and carry risks due to the necessity of sedation and anesthesia. Recently, several new approaches have been developed, including serological examination and magnetically controlled capsule endoscopy (MGCE). However, serological markers lack lesion information, while MGCE images lack molecular information. This study proposes combining these two technologies in a collaborative noninvasive diagnostic scheme as an alternative to the standard procedures. We introduce an interpretable framework for the clinical diagnosis of gastrointestinal diseases. Based on collected blood samples and MGCE records of patients with gastrointestinal diseases and comparisons with normal individuals, we selected serum metabolite signatures by bioinformatic analysis, captured image embedding signatures by convolutional neural networks, and inferred the location-specific associations between these signatures. Our study successfully identified five key metabolite signatures with functional relevance to gastrointestinal disease. The combined signatures achieved discrimination AUC of 0.88. Meanwhile, the image embedding signatures showed different levels of validation and testing accuracy ranging from 0.7 to 0.9 according to different locations in the gastrointestinal tract as explained by their specific associations with metabolite signatures. Overall, our work provides a new collaborative noninvasive identification pipeline and candidate metabolite biomarkers for image auxiliary diagnosis. This method should be valuable for the noninvasive detection and interpretation of gastrointestinal and other complex diseases.

Rewiring of Microbiota Networks in Erosive Inflammation of the Stomach and Small Bowel.

The development of non-invasive, inexpensive, and effective early diagnosis tests for gastric and small-bowel lesions is an urgent requirement. The introduction of magnetically guided capsule endoscopy (MGCE) has aided examination of the small bowel for diagnoses. However, the distribution of the fecal microbiome in abnormal erosions of the stomach and small bowel remains unclear. Herein, alternations in the fecal microbiome in three groups [normal, small-bowel inflammation, and chronic gastritis (CG)] were analyzed by metagenomics and our well-developed method [individual-specific edge-network analysis (iENA)]. In addition to the dominant microbiota identified by the conventional differential analysis, iENA could recognize novel network biomarkers of microbiome communities, such as the genus Bacteroide in CG and small-bowel inflammation. Combined with differential network analysis, the network-hub microbiota within rewired microbiota networks revealed high-ranked iENA microbiota markers, which were disease specific and had particular pathogenic functions. Our findings illuminate the components of the fecal microbiome and the importance of specific bacteria in CG and small-bowel erosions, and could be employed to develop preventive and non-invasive therapeutic strategies.

Integrative Analysis of Omics Big Data.

The diversity and huge omics data take biology and biomedicine research and application into a big data era, just like that popular in human society a decade ago. They are opening a new challenge from horizontal data ensemble (e.g., the similar types of data collected from different labs or companies) to vertical data ensemble (e.g., the different types of data collected for a group of person with match information), which requires the integrative analysis in biology and biomedicine and also asks for emergent development of data integration to address the great changes from previous population-guided to newly individual-guided investigations.Data integration is an effective concept to solve the complex problem or understand the complicate system. Several benchmark studies have revealed the heterogeneity and trade-off that existed in the analysis of omics data. Integrative analysis can combine and investigate many datasets in a cost-effective reproducible way. Current integration approaches on biological data have two modes: one is "bottom-up integration" mode with follow-up manual integration, and the other one is "top-down integration" mode with follow-up in silico integration.This paper will firstly summarize the combinatory analysis approaches to give candidate protocol on biological experiment design for effectively integrative study on genomics and then survey the data fusion approaches to give helpful instruction on computational model development for biological significance detection, which have also provided newly data resources and analysis tools to support the precision medicine dependent on the big biomedical data. Finally, the problems and future directions are highlighted for integrative analysis of omics big data.

Revisit of Machine Learning Supported Biological and Biomedical Studies.

Generally, machine learning includes many in silico methods to transform the principles underlying natural phenomenon to human understanding information, which aim to save human labor, to assist human judge, and to create human knowledge. It should have wide application potential in biological and biomedical studies, especially in the era of big biological data. To look through the application of machine learning along with biological development, this review provides wide cases to introduce the selection of machine learning methods in different practice scenarios involved in the whole biological and biomedical study cycle and further discusses the machine learning strategies for analyzing omics data in some cutting-edge biological studies. Finally, the notes on new challenges for machine learning due to small-sample high-dimension are summarized from the key points of sample unbalance, white box, and causality.