Pyroptosis burden is associated with anti-TNF treatment outcome in inflammatory bowel disease: new insights from bioinformatics analysis.

Biological agents known as anti-tumor necrosis factor (TNF) drugs are frequently utilized in the treatment of inflammatory bowel disease (IBD). In this study, we analyzed the shared processes of pyroptosis in Ulcerative colitis (UC) and Crohn's disease (CD), as well as explored the correlation between the burden of pyroptosis and the results of anti-TNF treatment based on bioinformatics analyses. We identified CAPS1, CASP5, GSDMD, AIM2, and NLRP3 as the hub genes, with AIM2 being the most effective indicator for predicting the response to anti-TNF therapy. We also noticed that non-responders received anti-TNF therapy exhibited elevated AIM2 protein expression. Subsequently, we conducted a cluster analysis based on AIM2-inflammasome-related genes and discovered that patients with a higher burden of AIM2 inflammasome displayed stronger immune function and a poor response to anti-TNF therapy. Overall, our study elucidates the pathway of pyroptosis in IBD and reveals AIM2 expression level as a potential biomarker for predicting the effectiveness of anti-TNF therapy.

Computation-Assisted Design of ssDNA Framework Nanorobots for Cancer Logical Recognition, Toehold Disintegration, Visual Dual-Diagnosis, and Synergistic Therapy.

Single-stranded DNA (ssDNA) allows flexible and directional modifications for multiple biological applications, while being greatly limited by their poor stability, increased folding errors, and complicated sequence optimizations. This greatly challenges the design and optimization of ssDNA sequences to fold stable 3D structures for diversified bioapplications. Herein, the stable pentahedral ssDNA framework nanorobots (ssDNA nanorobots) were intelligently designed, assisted by examining dynamic folding of ssDNA in self-assemblies via all-atom molecular dynamics simulations. Assisted by two functional siRNAs (S1 and S2), two ssDNA strands were successfully assembled into ssDNA nanorobots, which include five functional modules (skeleton fixation, logical dual recognition of tumor cell membrane proteins, enzyme loading, dual-miRNA detection and synergy siRNA loading) for multiple applications. By both theoretical calculations and experiments, ssDNA nanorobots were demonstrated to be stable, flexible, highly utilized with low folding errors. Thereafter, ssDNA nanorobots were successfully applied to logical dual-recognition targeting, efficient and cancer-selective internalization, visual dual-detection of miRNAs, selective siRNA delivery and synergistic gene silencing. This work has provided a computational pathway for constructing flexible and multifunctional ssDNA frameworks, enlarging biological application of nucleic acid nanostructures.

Biomineralization of DNA Nanoframeworks for Intracellular Delivery, On-Demand Diagnosis, and Synergistic Cancer Treatments.

DNA nanoframeworks, with great biological information and controlled framework structures, exhibit great potentials in biological applications. Their applications are normally limited by unstable structures susceptible to hydrolysis, depurination, depyrimidination, oxidation, alkylation, or nuclease degradations. Herein, to ensure the mechanical and chemical stabilities of DNA nanoframeworks for intracellular applications, biomineralization of multifunctional DNA nanoframeworks with a tetrahedral skeleton is employed. Via silicification, the S-S bond is simultaneously introduced to obtain the silica-armored DNA nanoframeworks (Si-DNA nanoframeworks), mechanically and chemically stabilized for efficient intracellular deliveries. This successfully prevents degradations and leakages of reagents loaded on Si-DNA nanoframeworks, including biomolecular siRNA and small DOX drugs. Furthermore, the nucleic acid strands of the nanoframeworks are labeled with FAM and the quencher, facilitating miRNA detection upon "turn-on" signals from hybridizations. Therefore, the nanoframeworks collapse via double responses of the silica coating (silica acidic dissolution and S-S reduction by GSH) in cancer cells, realizing on-demand reagent release for miRNA detection and synergistic treatments (by siRNA and DOX). Demonstrated by both in vivo and in vitro experiments, the biomineralization has stabilized DNA nanomaterials for biological applications.

Deep reinforcement learning for personalized treatment recommendation.

In precision medicine, the ultimate goal is to recommend the most effective treatment to an individual patient based on patient-specific molecular and clinical profiles, possibly high-dimensional. To advance cancer treatment, large-scale screenings of cancer cell lines against chemical compounds have been performed to help better understand the relationship between genomic features and drug response; existing machine learning approaches use exclusively supervised learning, including penalized regression and recommender systems. However, it would be more efficient to apply reinforcement learning to sequentially learn as data accrue, including selecting the most promising therapy for a patient given individual molecular and clinical features and then collecting and learning from the corresponding data. In this article, we propose a novel personalized ranking system called Proximal Policy Optimization Ranking (PPORank), which ranks the drugs based on their predicted effects per cell line (or patient) in the framework of deep reinforcement learning (DRL). Modeled as a Markov decision process, the proposed method learns to recommend the most suitable drugs sequentially and continuously over time. As a proof-of-concept, we conduct experiments on two large-scale cancer cell line data sets in addition to simulated data. The results demonstrate that the proposed DRL-based PPORank outperforms the state-of-the-art competitors based on supervised learning. Taken together, we conclude that novel methods in the framework of DRL have great potential for precision medicine and should be further studied.

The involvement of TNF-α and TNF-ß as proinflammatory cytokines in lymphocyte-mediated adaptive immunity of Nile tilapia by initiating apoptosis.

Tumor necrosis factors (TNFs) are pleiotropic cytokines with important functions in homeostasis and disease pathogenesis. Recent advances have shown that TNFs are also involved in the regulation of adaptive immune responses. However, the knowledge about how TNF participates in and regulates adaptive immune response in early vertebrates is still limited. In present study, we identified two isoforms of TNF, TNF-α and TNF-ß, from Nile tilapia Oreochromis niloticus (On-TNF-α and ß). After analyzing the sequence characteristics, we investigated their regulatory roles in adaptive immune response of this fish species. On-TNF-α and ß are evolutionarily conserved compare with their homologs from other vertebrates. Both TNFs were distributed in a wide range of tissues in O. niloticus, and with relative higher expression level in gill. After the animals were infected by Streptococcus agalactiae, mRNA levels of On-TNF-α and TNF-ß in spleen lymphocytes were significantly upregulated during the primary response stage of adaptive immunity. Meanwhile, both TNF proteins in spleen lymphocytes were also dramatically elevated during the adaptive immune stage after bacterial infection. These results indicate the potential participation of On-TNF-α and TNF-ß in adaptive immune response of Nile tilapia. Furthermore, On-TNF-α and ß transcripts were obviously augmented, once spleen lymphocytes were activated by T cell-specific mitogen PHA. More importantly, both recombinant On-TNF-α and ß could induce the apoptosis of head-kidney leukocytes of Nile tilapia. And On-TNF-ß but not On-TNF-α promoted the apoptosis by activating caspase-8 in the target cells. Altogether, our study revealed that TNF-α and TNF-ß participated in the lymphocyte-mediated adaptive immune response of Nile tilapia by initiating the apoptosis, and thus shed novel perspective for the regulatory mechanism of adaptive immunity in teleost.

Metal-DNA coordination based bioinspired hybrid nanospheres for in situ amplification and sensing of microRNA.

Sufficient delivery of biomolecules into cells with high loading efficiency and easy cleavability would be significant for the visualization of biomolecules in living cells. Herein, a facile approach based on nano-wire balls (NWs) for efficient loading, intracellular delivery of nucleic acids and in situ targeted miRNA bioimaging is proposed, by feeding of Zn ions for generating DNA-inorganic hybrid structures with large surface areas and good stability. Given that the versatile and robust hybridization chain reaction (HCR) amplification strategy combines DNA assembly with intracellular assay, the resulting NWs without any complicated modification are capable of enhanced signals for the targeted imaging of cancer cells. This method realized a linear detection range of 100 fM to 10 nM, with a low detection limit of 83.6 fM in vitro, and could be used to effectively differentiate the expression levels of miRNA-21 in living cells. Due to its high loading efficiency, excellent biocompatibility and low toxicity, this system can be used to construct a coordination-based delivery nanoplatform for in situ enzyme-free amplified imaging of miRNAs, expanding the application of DNA-based nanomaterials for cellular delivery and intracellular molecule analysis.

S6K1/S6 axis-regulated lymphocyte activation is important for adaptive immune response of Nile tilapia.

Ribosomal protein S6 kinase beta-1 (S6K1) is a serine/threonine kinase downstream of the mechanistic target of rapamycin (mTOR) pathway, and plays crucial roles in immune regulation. Although remarkable progress has been achieved with a mouse model, how S6K1 regulates adaptive immunity is largely unknown in early vertebrates. In this study, we identified an S6K1 from Nile tilapia Oreochromis niloticus (OnS6K1), and further investigated its potential regulatory role on the adaptive immunity of this fish species. Both sequence and structure of OnS6K1 were highly conserved with its homologs from other vertebrates and invertebrates. OnS6K1 was widely expressed in immune tissues, and with a relative higher expression level in the liver, spleen and head kidney. At the adaptive immune stage of Nile tilapia that infected with Aeromonas hydrophila, mRNA expression of OnS6K1 and its downstream effector S6 was significantly up-regulated in spleen lymphocytes. Meanwhile, their phosphorylation level was also enhanced during this process, suggesting that S6K1/S6 axis participated in the primary response of anti-bacterial adaptive immunity in Nile tilapia. Furthermore, after spleen lymphocytes were activated by the T cell-specific mitogen PHA or lymphocytes agonist PMA in vitro, mRNA and phosphorylation levels of S6K1 were elevated, and phosphorylation of S6 was also enhanced. Once S6K1 activity was blocked by a specific inhibitor, both mRNA and phosphorylation levels of S6 were severely impaired. More importantly, blockade of S6K1/S6 axis reduced the expression of T cell activation marker IFN-γ and CD122 in PHA-activated spleen lymphocytes, indicating the essential role of S6K1/S6 axis in regulating T cell activation of Nile tilapia. Together, our study suggests that S6K1 and its effector S6 regulate lymphocyte activation of Nile tilapia, and in turn promote lymphocyte-mediated adaptive immunity. This study enriched the mechanism of adaptive immune response in teleost and provided useful clues to understand the evolution of adaptive immune system.

Chemiluminescence Resonance Energy Transfer-Based Mesoporous Silica Nanosensors for the Detection of miRNA.

The chemiluminescence resonance energy transfer (CRET)-based method is free of autofluorescence interference, which can achieve an extremely high signal-to-background ratio for detection. Nevertheless, this method is still hindered by the inner filter effect, quenching effect, and nonspecific absorption of reported nanoparticles. Herein, mesoporous silica nanomaterials (MSNs) acted as carriers to load both the donor (horseradish peroxidase, HRP) and the acceptor (a functional DNA duplex). This approach realized the construction of a new CRET-based nanosensor for the sensitive detection of miRNA. By controlling the energy-transfer distance with the designed DNAs, the donor emission at 430 nm could be quenched by the adsorption of the dye labeled on the acceptor DNA. The CRET system could be destroyed by releasing acceptor DNA from linker DNA via the competitive hybridization of target miRNA, resulting in emission recovery for quantification. With the cancer biomarker miR-155 as the model, the sensitive and selective detection of miR-155 was achieved, which showed high energy-transfer efficiency, good specificity, favorable biodegradability, and low toxicity. This work provides a potential pathway for biological detection and clinical diagnosis.

Multi-Dimensionally Extended Functionalization Innovates to an Entropy-Driven Detection of Multi-miRNAs for One-Step Cancer Screening and Diagnosis in Living Cells.

Compared with tedious multi-step detections, multi-functional nanoprobes are effective for one-step screening and diagnosis of cancers by multi-detection of microRNAs (miRNAs). However, limited probe density, spatial mutual interference, and low target-triggered hybridization efficiency of nanoprobes will hinder intracellular applications. Here, for obtaining high loading density but low spatial mutual interference between functional biomolecules on nanoprobes, an extended biofunctionalization in three dimensions (the two-dimensional surface and a special "height" direction) is designed. Therefore, a multi-functional probe is constructed for one-step detection of multi-miRNAs for cancer screening and diagnosis. With linker-bridged multiple single-stranded DNAs swung out rigidly, multi-dimensionally extended upconversion nanorods (ME-UCNRs) covered by chitosan are constructed to load and deliver multiple biomolecules into living cells. Escaping from endolysosomes, ME-UCNRs maintain good biological activities of functionalized DNAs for effective detection of multi-miRNAs in living cells. Thereby, with multiple targets of miRNAs, toehold-mediated entropy-driven strand displacements are employed to give respectively changed fluorescent signals via fluorescence resonance energy transfer. Thus, a universal cancer biomarker of miR-21 and two specific liver-cancer biomarkers (miR-199a and miR-224) are efficiently detected through a one-step detection. By discriminating cancer cells from normal ones and determining liver-cancer cells simultaneously, this work innovates an efficient and definite one-step strategy for fast screening and early cancer diagnosis.

Outcome weighted ψ-learning for individualized treatment rules.

An individualized treatment rule is often employed to maximize a certain patient-specific clinical outcome based on his/her clinical or genomic characteristics as well as heterogeneous response to treatments. Although developing such a rule is conceptually important to personalized medicine, existing methods such as the partial least squares Qian and Murphy (2011) suffers from the difficulty of indirect maximization of a patient's clinical outcome, while the outcome weighted learning Y. Zhao, Zeng, Rush, and Kosorok (2012) is not robust against any perturbation of the outcome. In this article, we propose a weighted ψ-learning method to optimize an individualized treatment rule, which is robust against any data perturbation near the decision boundary by seeking the maximum separation. To solve nonconvex minimization, we employ a difference convex algorithm to relax the non-convex minimization iteratively based on a decomposition of the cost function into a difference of two convex functions. On this ground, we also introduce a variable selection method for further removing redundant variables for a higher performance. Finally, we illustrate the proposed method by simulations and a lung health study and demonstrate that it yields higher performances in terms of accuracy of prediction of individualized treatment.

Biodegradable nanosyringes for intracellular amplification-based dual-diagnosis and gene therapy in single living cells.

The efficient delivery of biomolecules into living cells as well as their easy biodegradation have been challenges for the application of intracellular amplification for sensitive multiple-diagnosis and gene therapy for cancer. Herein, new strategies of amplification-based dual-detection of cancer biomarkers (Let-7a miRNA and VEGF) and gene therapy for cancers are put forward. These are achieved through biodegradable nanosyringes (NSs), rigid and sharp in vitro but degradable in vivo, which are applied for the efficient loading, delivery and release of biomolecules (enzymes, nucleic acids, and even silencing RNA) into living cells. After penetrating cell membranes and escaping from endosomes through their rigid and sharp tips, NSs release biomolecules for fast and easy "one-step" rolling circle amplification (ring formation and amplification) in single living cells. Therefore, based on signals from two probes, FAM-Probe and Cy5-Probe, that selectively bind to amplification products, 100 aM of Let-7a and 100 fM of VEGF could be detected, which are much lower than reported values. Furthermore, siRNAs can also be delivered by NSs for gene therapy, and their therapeutic effect was evaluated by their in vivo antitumor efficacy in CCRF-CEM subcutaneous xenograft nude mice. Rigid in vitro and degradable in vivo, NSs show potential for achieving fast, sensitive and safe cancer diagnosis and efficient therapy.

The Improvement and Application of Lentivirus-Mediated Gene Transfer and Expression System in Penaeid Shrimp Cells.

This study first reported the improvement and application of lentivirus-mediated gene transfer and expression system in shrimp cells. After modified by the inclusion of two envelope proteins (VP19 and VP28) of shrimp white spot syndrome virus (WSSV) into the envelope of the packaged lentivirus, and insertion of a truncated promoter of immediate-early gene 1 (Pie1-504) of shrimp WSSV virus into the lentiviral reporter plasmid, the second-generation lentiviral expression system (pLVX-PEF1α-IRES-mCherry, psPAX2, and PMD2.G) was found to behave better in the mitosis-arrested shrimp cells than the similarly modified retrovirus expression system did. Results from the insect sf9 cells indicated that the inclusion of VP19 and VP28 into the envelope of packaged lentiviruses could significantly improve the tropism or infectivity of the modified lentiviruses to insect cells in a cumulative way. Notably, the VP28 contributed about 86% of the total increase of the tropism. In the shrimp primary lymphoid cells infected by modified lentivirus IV with both VP19 and VP28 included, the infection efficiency was up to 11% (non-confocal) and 19% (confocal) and no background fluorescent signal was observed. However, background fluorescent signal was observed in the shrimp primary Oka organ cells although only under a confocal microscope. In the lentivirus IV-infected Oka organ cells, the actual infection efficiencies were calculated up to 8% (non-confocal) and 19% (confocal), significantly higher than those of commercial intact lentivirus I of 0 (non-confocal) and 3% (confocal). The insertion of WSSV promoter (Pie1-504) had interrupted the effective expression of reporter plasmid encoding lentiviral construct of pLVX-PEF1α-Pie1-504-IRES-mCherry in the HEK293T cells, but markedly increased its efficiencies up to 14% (non-confocal) and 26% (confocal) in the Oka organ cells. This improved lentivirus expression system will provide us a useful tool for efficient gene transfer and expression in shrimp cells.

Sandwich DNA Hybridization Fluorescence Resonance Energy-Transfer Strategy for miR-122 Detection by Core-Shell Upconversion Nanoparticles.

An upconversion nanoparticle (UCNP)-based fluorescence resonance energy-transfer (FRET) strategy is normally restricted by the complicated preparations, low energy-transfer efficiency, and the challenge on improving specificity. Herein, simple DNA-functionalized UCNPs were designed as energy donors for constructing a FRET-based probe to detect the liver-specific microRNA 122 (miR-122). To improve FRET efficiency, UCNPs were constructed with confined core-shell structures, in which emitting ions were precisely located in the thin shell to make them close enough to external energy acceptors. Subsequently, capture DNA was simply functionalized on the outer surface of UCNPs based on ligand exchange that contributed to shortening the energy-transfer distance without extra modification. To gain high specificity, the donor-to-acceptor distance of FRET was controlled by a sandwich DNA hybridization structure using two shorter DNAs with designed complementary sequences (capture DNA and dye-labeled report DNA) to capture the longer target of miR-122. Therefore, the sensitive detection of miR-122 was achieved based on the decreased signals of UCNPs and the increased signals of the dye labeled on reported DNA. With good biocompatibility, this method has been further applied to cancer cell imaging and in vivo imaging, which opened up a new avenue to the sensitive detection and imaging of microRNA in biological systems.

A NOVEL AND EFFICIENT ALGORITHM FOR DE NOVO DISCOVERY OF MUTATED DRIVER PATHWAYS IN CANCER.

Next-generation sequencing studies on cancer somatic mutations have discovered that driver mutations tend to appear in most tumor samples, but they barely overlap in any single tumor sample, presumably because a single driver mutation can perturb the whole pathway. Based on the corresponding new concepts of coverage and mutual exclusivity, new methods can be designed for de novo discovery of mutated driver pathways in cancer. Since the computational problem is a combinatorial optimization with an objective function involving a discontinuous indicator function in high dimension, many existing optimization algorithms, such as a brute force enumeration, gradient descent and Newton's methods, are practically infeasible or directly inapplicable. We develop a new algorithm based on a novel formulation of the problem as non-convex programming and non-convex regularization. The method is computationally more efficient, effective and scalable than existing Monte Carlo searching and several other algorithms, which have been applied to The Cancer Genome Atlas (TCGA) project. We also extend the new method for integrative analysis of both mutation and gene expression data. We demonstrate the promising performance of the new methods with applications to three cancer datasets to discover de novo mutated driver pathways.

Advanced Interfere Treatment of Diabetic Cardiomyopathy Rats by aFGF-Loaded Heparin-Modified Microbubbles and UTMD Technique.

This study aims to investigate the preclinical performance and mechanism of a novel strategy of aFGF-loaded heparin-modified microbubbles (aFGF-HMB) combined with ultrasound-targeted microbubble destruction (UTMD) technique for diabetic cardiomyopathy (DCM) prevention. Type 1 diabetic rats were induced by streptozotocin. Twelve weeks after intervention, indexes from transthoracic echocardiography and cardiac catheterization showed that the left ventricular function in the aFGF-HMB/UTMD group was significantly improved compared with diabetes control (DM). From Picrosirius Red staining and TUNEL staining, the aFGF-HMB/UTMD group showed significant difference from the other groups. The cardiac collagen volume fraction (CVF) and myocardial cell apoptosis index (AI) in aFGF-HMB/UTMD group decreased to 7.2 % and 7.11 % respectively, compared with the DM group (CVF = 24.5 % and AI =20.3 % respectively). The results of myocardial microvascular density (MCD) also proved the strongest inhibition of aFGF-HMB/UTMD group on DCM progress. CD31 staining of aFGF-HMB/UTMD group reached 22 n/hrp, much higher than that of DM group (9 n/hrp). These results confirmed that the abnormalities including left ventricular dysfunction, myocardial fibrosis, cardiomyocytes apoptosis and microvascular rarefaction could be suppressed by twice weekly aFGF treatments for 12 consecutive weeks (free aFGF or aFGF-HMB+/-UTMD), with the strongest improvements observed in the aFGF-HMB/UTMD group (P < 0.05 vs free aFGF or aFGF-HMB). Western blot analyses of heart tissue further revealed the highest aFGF, anti-apoptosis protein (Bcl-2), VEGF-C, pAkt, pFoxo-3a levels and strongest reduction in pro-apoptosis proteins (Bax) level in aFGF-HMB/UTMD group. Overall, aFGF-HMB combined with UTMD technique might be developed as an effective strategy to prevent DCM in future clinical therapy.

Integrative and regularized principal component analysis of multiple sources of data.

Integration of data of disparate types has become increasingly important to enhancing the power for new discoveries by combining complementary strengths of multiple types of data. One application is to uncover tumor subtypes in human cancer research in which multiple types of genomic data are integrated, including gene expression, DNA copy number, and DNA methylation data. In spite of their successes, existing approaches based on joint latent variable models require stringent distributional assumptions and may suffer from unbalanced scales (or units) of different types of data and non-scalability of the corresponding algorithms. In this paper, we propose an alternative based on integrative and regularized principal component analysis, which is distribution-free, computationally efficient, and robust against unbalanced scales. The new method performs dimension reduction simultaneously on multiple types of data, seeking data-adaptive sparsity and scaling. As a result, in addition to feature selection for each type of data, integrative clustering is achieved. Numerically, the proposed method compares favorably against its competitors in terms of accuracy (in identifying hidden clusters), computational efficiency, and robustness against unbalanced scales. In particular, compared with a popular method, the new method was competitive in identifying tumor subtypes associated with distinct patient survival patterns when applied to a combined analysis of DNA copy number, mRNA expression, and DNA methylation data in a glioblastoma multiforme study. Copyright © 2016 John Wiley & Sons, Ltd.

Glioma-targeted therapy using Cilengitide nanoparticles combined with UTMD enhanced delivery.

Malignant gliomas especially glioblastoma (GBM) are poorly responsive to the current treatments. Cilengitide (CGT) is a cyclic pentapeptide that demonstrated efficacy for GBM treatment by targeting the integrins avß3 and avß5 over-expressed on GBM cells. However, clinical translation of this therapy has been limited by issues including fast blood clearance, high kidney and liver uptake, poor blood-brain barrier (BBB) penetration, low tumor specificity and rapid washout from tumors. In this study, these issues were tackled in an integrated manner using a multi-stage strategy combining ultrasound-targeted microbubble destruction (UTMD) with CGT nanotherapy. CGT nanoparticles (CGT-NP) prepared using gelatin and Poloxamer 188-grafted heparin copolymer demonstrated significant apoptotic and cytotoxic effects in C6 GBM cells. Biodistribution study in a rat GBM model demonstrated buildup of high CGT level in tumors subjected to CGT-NP+UTMD combined therapy. The tumor CGT level in these animals was increased over 3-fold, tumor retention of CGT prolonged and renal clearance significantly reduced when compared with free CGT with or without UTMD. CGT-NP+UTMD treatment was further shown to extend the median survival period from less than 20days in the control and about 30days in free CGT group to about 80days. This was achieved with low CGT dosing level (2mg/kg twice weekly). In situ monitoring of GFAP, Ki67, caspase-3, Beclin-1, and LC-3 in the tumor samples together with TUNEL assay, transmission electron microscope imaging and Western blot assay all demonstrated high apoptotic and autophagy activities induced by the combined therapy. In conclusion, this study has provided extensive preclinical data supporting the use of this combined therapy to overcome the limitations of standard CGT treatment of gliomas.

Estimation of multiple networks in Gaussian mixture models.

We aim to estimate multiple networks in the presence of sample heterogeneity, where the independent samples (i.e. observations) may come from different and unknown populations or distributions. Specifically, we consider penalized estimation of multiple precision matrices in the framework of a Gaussian mixture model. A major innovation is to take advantage of the commonalities across the multiple precision matrices through possibly nonconvex fusion regularization, which for example makes it possible to achieve simultaneous discovery of unknown disease subtypes and detection of differential gene (dys)regulations in functional genomics. We embed in the EM algorithm one of two recently proposed methods for estimating multiple precision matrices in Gaussian graphical models. We demonstrate the feasibility and potential usefulness of the proposed methods in an application to glioblastoma subtype discovery and differential gene network analysis with a microarray gene expression data set. We also conduct realistic simulation studies to evaluate and compare the performance of various methods.

Nonlinear Joint Latent Variable Models and Integrative Tumor Subtype Discovery.

Integrative analysis has been used to identify clusters by integrating data of disparate types, such as deoxyribonucleic acid (DNA) copy number alterations and DNA methylation changes for discovering novel subtypes of tumors. Most existing integrative analysis methods are based on joint latent variable models, which are generally divided into two classes: joint factor analysis and joint mixture modeling, with continuous and discrete parameterizations of the latent variables respectively. Despite recent progresses, many issues remain. In particular, existing integration methods based on joint factor analysis may be inadequate to model multiple clusters due to the unimodality of the assumed Gaussian distribution, while those based on joint mixture modeling may not have the ability for dimension reduction and/or feature selection. In this paper, we employ a nonlinear joint latent variable model to allow for flexible modeling that can account for multiple clusters as well as conduct dimension reduction and feature selection. We propose a method, called integrative and regularized generative topographic mapping (irGTM), to perform simultaneous dimension reduction across multiple types of data while achieving feature selection separately for each data type. Simulations are performed to examine the operating characteristics of the methods, in which the proposed method compares favorably against the popular iCluster that is based on a linear joint latent variable model. Finally, a glioblastoma multiforme (GBM) dataset is examined.

Structural pursuit over multiple undirected graphs.

Gaussian graphical models are useful to analyze and visualize conditional dependence relationships between interacting units. Motivated from network analysis under di erent experimental conditions, such as gene networks for disparate cancer subtypes, we model structural changes over multiple networks with possible heterogeneities. In particular, we estimate multiple precision matrices describing dependencies among interacting units through maximum penalized likelihood. Of particular interest are homogeneous groups of similar entries across and zero-entries of these matrices, referred to as clustering and sparseness structures, respectively. A non-convex method is proposed to seek a sparse representation for each matrix and identify clusters of the entries across the matrices. Computationally, we develop an e cient method on the basis of di erence convex programming, the augmented Lagrangian method and the block-wise coordinate descent method, which is scalable to hundreds of graphs of thousands nodes through a simple necessary and sufficient partition rule, which divides nodes into smaller disjoint subproblems excluding zero-coe cients nodes for arbitrary graphs with convex relaxation. Theoretically, a finite-sample error bound is derived for the proposed method to reconstruct the clustering and sparseness structures. This leads to consistent reconstruction of these two structures simultaneously, permitting the number of unknown parameters to be exponential in the sample size, and yielding the optimal performance of the oracle estimator as if the true structures were given a priori. Simulation studies suggest that the method enjoys the benefit of pursuing these two disparate kinds of structures, and compares favorably against its convex counterpart in the accuracy of structure pursuit and parameter estimation.