Identification of Key Genes for the Ultrahigh Yield of Rice Using Dynamic Cross-tissue Network Analysis

Integrating multi-omics data of childhood asthma using a deep association model

Childhood asthma is one of the most common respiratory diseases with rising mortality and morbidity. The multi-omics data is providing a new chance to explore collaborative biomarkers and corresponding diagnostic models of childhood asthma. To capture the nonlinear association of multi-omics data and improve interpretability of diagnostic model, we proposed a novel deep association model (DAM) and corresponding efficient analysis framework. First, the Deep Subspace Reconstruction was used to fuse the omics data and diagnostic information, thereby correcting the distribution of the original omics data and reducing the influence of unnecessary data noises. Second, the Joint Deep Semi-Negative Matrix Factorization was applied to identify different latent sample patterns and extract biomarkers from different omics data levels. Third, our newly proposed Deep Orthogonal Canonical Correlation Analysis can rank features in the collaborative module, which are able to construct the diagnostic model considering nonlinear correlation between different omics data levels. Using DAM, we deeply analyzed the transcriptome and methylation data of childhood asthma. The effectiveness of DAM is verified from the perspectives of algorithm performance and biological significance on the independent test dataset, by ablation experiment and comparison with many baseline methods from clinical and biological studies. The DAM-induced diagnostic model can achieve a prediction AUC of 0.912, which is higher than that of many other alternative methods. Meanwhile, relevant pathways and biomarkers of childhood asthma are also recognized to be collectively altered on the gene expression and methylation levels. As an interpretable machine learning approach, DAM simultaneously considers the non-linear associations among samples and those among biological features, which should help explore interpretative biomarker candidates and efficient diagnostic models from multi-omics data analysis for human complex diseases.

Identification of STAT3 as a biomarker for cellular senescence in liver fibrosis: A bioinformatics and experimental validation study

BACKGROUND

Cellular senescence is associated with a dysregulated inflammatory response, which is an important driver of the development of liver fibrosis (LF). This study aimed to investigate the effect of cellular senescence on LF and identify potential key biomarkers through bioinformatics analysis combined with validation experiments in vivo and in vitro.

METHODS

The Gene Expression Omnibus (GEO) database and GeneCards database were used to download the LF dataset and the aging-related gene set, respectively. Functional enrichment analysis of differential genes was then performed using GO and KEGG. Hub genes were further screened using Cytoscape's cytoHubba. Diagnostic values for hub genes were evaluated with a receiver operating characteristic (ROC) curve. Next, CIBERSORTx was used to estimate immune cell types and ratios. Finally, in vivo and in vitro experiments validated the results of the bioinformatics analysis. Moreover, molecular docking was used to simulate drug-gene interactions.

RESULTS

A total of 44 aging-related differentially expressed genes (AgDEGs) were identified, and enrichment analysis showed that these genes were mainly enriched in inflammatory and immune responses. PPI network analysis identified 6 hub AgDEGs (STAT3, TNF, MMP9, CD44, TGFB1, and TIMP1), and ROC analysis showed that they all have good diagnostic value. Immune infiltration suggested that hub AgDEGs were significantly associated with M1 macrophages or other immune cells. Notably, STAT3 was positively correlated with α-SMA, COL1A1, IL-6 and IL-1β, and was mainly expressed in hepatocytes (HCs). Validation experiments showed that STAT3 expression was upregulated and cellular senescence was increased in LF mice. A co-culture system of HCs and hepatic stellate cells (HSCs) further revealed that inhibiting STAT3 reduced HCs senescence and suppressed HSCs activation. In addition, molecular docking revealed that STAT3 was a potential drug therapy target.

CONCLUSIONS

STAT3 may be involved in HCs senescence and promote HSCs activation, which in turn leads to the development of LF. Our findings suggest that STAT3 could be a potential biomarker for LF.

Integration of multi-omics technologies for crop improvement: Status and prospects

With the rapid development of next-generation sequencing (NGS), multi-omics techniques have been emerging as effective approaches for crop improvement. Here, we focus mainly on addressing the current status and future perspectives toward omics-related technologies and bioinformatic resources with potential applications in crop breeding. Using a large amount of omics-level data from the functional genome, transcriptome, proteome, epigenome, metabolome, and microbiome, clarifying the interaction between gene and phenotype formation will become possible. The integration of multi-omics datasets with pan-omics platforms and systems biology could predict the complex traits of crops and elucidate the regulatory networks for genetic improvement. Different scales of trait predictions and decision-making models will facilitate crop breeding more intelligent. Potential challenges that integrate the multi-omics data with studies of gene function and their network to efficiently select desirable agronomic traits are discussed by proposing some cutting-edge breeding strategies for crop improvement. Multi-omics-integrated approaches together with other artificial intelligence techniques will contribute to broadening and deepening our knowledge of crop precision breeding, resulting in speeding up the breeding process.

Interpreting Functional Impact of Genetic Variations by Network QTL for Genotype–Phenotype Association Study

An enormous challenge in the post-genome era is to annotate and resolve the consequences of genetic variation on diverse phenotypes. The genome-wide association study (GWAS) is a well-known method to identify potential genetic loci for complex traits from huge genetic variations, following which it is crucial to identify expression quantitative trait loci (eQTL). However, the conventional eQTL methods usually disregard the systematical role of single-nucleotide polymorphisms (SNPs) or genes, thereby overlooking many network-associated phenotypic determinates. Such a problem motivates us to recognize the network-based quantitative trait loci (QTL), i.e., network QTL (nQTL), which is to detect the cascade association as genotype → network → phenotype rather than conventional genotype → expression → phenotype in eQTL. Specifically, we develop the nQTL framework on the theory and approach of single-sample networks, which can identify not only network traits (e.g., the gene subnetwork associated with genotype) for analyzing complex biological processes but also network signatures (e.g., the interactive gene biomarker candidates screened from network traits) for characterizing targeted phenotype and corresponding subtypes. Our results show that the nQTL framework can efficiently capture associations between SNPs and network traits (i.e., edge traits) in various simulated data scenarios, compared with traditional eQTL methods. Furthermore, we have carried out nQTL analysis on diverse biological and biomedical datasets. Our analysis is effective in detecting network traits for various biological problems and can discover many network signatures for discriminating phenotypes, which can help interpret the influence of nQTL on disease subtyping, disease prognosis, drug response, and pathogen factor association. Particularly, in contrast to the conventional approaches, the nQTL framework could also identify many network traits from human bulk expression data, validated by matched single-cell RNA-seq data in an independent or unsupervised manner. All these results strongly support that nQTL and its detection framework can simultaneously explore the global genotype–network–phenotype associations and the underlying network traits or network signatures with functional impact and importance.

Deep latent space fusion for adaptive representation of heterogeneous multi-omics data

The integration of multi-omics data makes it possible to understand complex biological organisms at the system level. Numerous integration approaches have been developed by assuming a common underlying data space. Due to the noise and heterogeneity of biological data, the performance of these approaches is greatly affected. In this work, we propose a novel deep neural network architecture, named Deep Latent Space Fusion (DLSF), which integrates the multi-omics data by learning consistent manifold in the sample latent space for disease subtypes identification. DLSF is built upon a cycle autoencoder with a shared self-expressive layer, which can naturally and adaptively merge nonlinear features at each omics level into one unified sample manifold and produce adaptive representation of heterogeneous samples at the multi-omics level. We have assessed DLSF on various biological and biomedical datasets to validate its effectiveness. DLSF can efficiently and accurately capture the intrinsic manifold of the sample structures or sample clusters compared with other state-of-the-art methods, and DLSF yielded more significant outcomes for biological significance, survival prognosis and clinical relevance in application of cancer study in The Cancer Genome Atlas. Notably, as a deep case study, we determined a new molecular subtype of kidney renal clear cell carcinoma that may benefit immunotherapy in the viewpoint of multi-omics, and we further found potential subtype-specific biomarkers from multiple omics data, which were validated by independent datasets. In addition, we applied DLSF to identify potential therapeutic agents of different molecular subtypes of chronic lymphocytic leukemia, demonstrating the scalability of DLSF in diverse omics data types and application scenarios.

Landscape dynamic network biomarker analysis reveals the tipping point of transcriptome reprogramming to prevent skin photodamage

Skin, as the outmost layer of human body, is frequently exposed to environmental stressors including pollutants and ultraviolet (UV), which could lead to skin disorders. Generally, skin response process to ultraviolet B (UVB) irradiation is a nonlinear dynamic process, with unknown underlying molecular mechanism of critical transition. Here, the landscape dynamic network biomarker (l-DNB) analysis of time series transcriptome data on 3D skin model was conducted to reveal the complicated process of skin response to UV irradiation at both molecular and network levels. The advanced l-DNB analysis approach showed that: (i) there was a tipping point before critical transition state during pigmentation process, validated by 3D skin model; (ii) 13 core DNB genes were identified to detect the tipping point as a network biomarker, supported by computational assessment; (iii) core DNB genes such as COL7A1 and CTNNB1 can effectively predict skin lightening, validated by independent human skin data. Overall, this study provides new insights for skin response to repetitive UVB irradiation, including dynamic pathway pattern, biphasic response, and DNBs for skin lightening change, and enables us to further understand the skin resilience process after external stress.

The Elite Alleles of OsSPL4 Regulate Grain Size and Increase Grain Yield in Rice

Grain weight and grain number, the two important yield traits, are mainly determined by grain size and panicle architecture in rice. Herein, we report the identification and functional analysis of OsSPL4 in panicle and grain development of rice. Using CRISPR/Cas9 system, two elite alleles of OsSPL4 were obtained, which exhibited an increasing number of grains per panicle and grain size, resulting in increase of rice yield. Cytological analysis showed that OsSPL4 could regulate spikelet development by promoting cell division. The results of RNA-seq and qRT-PCR validations also demonstrated that several MADS-box and cell-cycle genes were up-regulated in the mutation lines. Co-expression network revealed that many yield-related genes were involved in the regulation network of OsSPL4. In addition, OsSPL4 could be cleaved by the osa-miR156 in vivo, and the OsmiR156-OsSPL4 module might regulate the grain size in rice. Further analysis indicated that the large-grain allele of OsSPL4 in indica rice might introgress from aus varieties under artificial selection. Taken together, our findings suggested that OsSPL4 could be as a key regulator of grain size by acting on cell division control and provided a strategy for panicle architecture and grain size modification for yield improvement in rice.

OsSPL9 Regulates Grain Number and Grain Yield in Rice

Rice grain yield consists of several key components, including tiller number, grain number per panicle (GNP), and grain weight. Among them, GNP is mainly determined by panicle branches and spikelet formation. In this study, we identified a gene affecting GNP and grain yield, OsSPL9, which encodes SQUAMOSA-PROMOTER BINDING PROTEIN-LIKE (SPL) family proteins. The mutation of OsSPL9 significantly reduced secondary branches and GNP. OsSPL9 was highly expressed in the early developing young panicles, consistent with its function of regulating panicle development. By combining expression analysis and dual-luciferase assays, we further confirmed that OsSPL9 directly activates the expression of RCN1 (rice TERMINAL FLOWER 1/CENTRORADIALIS homolog) in the early developing young panicle to regulate the panicle branches and GNP. Haplotype analysis showed that Hap3 and Hap4 of OsSPL9 might be favorable haplotypes contributing to high GNP in rice. These results provide new insights on high grain number breeding in rice.