Spatially contrastive variational autoencoder for deciphering tissue heterogeneity from spatially resolved transcriptomics.

Recent advances in spatially resolved transcriptomics (SRT) have brought ever-increasing opportunities to characterize expression landscape in the context of tissue spatiality. Nevertheless, there still exist multiple challenges to accurately detect spatial functional regions in tissue. Here, we present a novel contrastive learning framework, SPAtially Contrastive variational AutoEncoder (SpaCAE), which contrasts transcriptomic signals of each spot and its spatial neighbors to achieve fine-grained tissue structures detection. By employing a graph embedding variational autoencoder and incorporating a deep contrastive strategy, SpaCAE achieves a balance between spatial local information and global information of expression, enabling effective learning of representations with spatial constraints. Particularly, SpaCAE provides a graph deconvolutional decoder to address the smoothing effect of local spatial structure on expression's self-supervised learning, an aspect often overlooked by current graph neural networks. We demonstrated that SpaCAE could achieve effective performance on SRT data generated from multiple technologies for spatial domains identification and data denoising, making it a remarkable tool to obtain novel insights from SRT studies.

Development of therapeutic monoclonal antibodies against DKK1 peptide-HLA-A2 complex to treat human cancers.

BACKGROUND: Targeted immunotherapy with monoclonal antibodies (mAbs) is an effective and safe method for the treatment of malignancies. Development of mAbs with improved cytotoxicity, targeting new and known tumor-associated antigens, therefore continues to be an active research area. We reported that Dickkopf-1 (DKK1) is a good target for immunotherapy of human cancers based on its wide expression in different cancers but not in normal tissues. As DKK1 is a secreted protein, mAbs binding directly to DKK1 have limited effects on cancer cells in vivo. METHODS: The specificity and antibody-binding capacity of DKK1-A2 mAbs were determined using indirect ELISA, confocal imaging, QIFIKIT antibody-binding capacity and cell surface binding assays. The affinity of mAbs was determined using a surface plasmon resonance biosensor. A flow cytometry-based cell death was performed to detect tumor cell apoptosis. Antibody-dependent cellular cytotoxicity (ADCC) and complement-dependent cytotoxicity (CDC) assays were used to evaluate the ability of DKK1-A2 mAbs to mediate ADCC and CDC activities against tumor cells in vitro. Flow cytometry data were collected with an FACSymphony A3 cell analyzer and analyzed with FlowJo V.10.1 software. Human cancer xenograft mouse models were used to determine the in vivo therapeutic efficacy and the potential safety and toxicity of DKK1-A2 mAbs. In situ TUNEL assay was performed to detect apoptosis in tumors and mouse organs. RESULTS: We generated novel DKK1-A2 mAbs that recognize the DKK1 P20 peptide presented by human HLA-A*0201 (HLA-A2) molecules (DKK1-A2 complexes) that are naturally expressed by HLA-A2+DKK1+ cancer cells. These mAbs directly induced apoptosis in HLA-A2+DKK1+ hematologic and solid cancer cells by activating the caspase-9 cascade, effectively lysed the cancer cells in vitro by mediating CDC and ADCC and were therapeutic against established cancers in their xenograft mouse models. As DKK1 is not detected in most human tissues, DKK1-A2 mAbs neither bound to or killed HLA-A2+ blood cells in vitro nor caused tissue damage in tumor-free or tumor-bearing HLA-A2-transgenic mice. CONCLUSION: Our study suggests that DKK1-A2 mAbs may be a promising therapeutic agent to treat human cancers.

Contrastively generative self-expression model for single-cell and spatial multimodal data.

Advances in single-cell multi-omics technology provide an unprecedented opportunity to fully understand cellular heterogeneity. However, integrating omics data from multiple modalities is challenging due to the individual characteristics of each measurement. Here, to solve such a problem, we propose a contrastive and generative deep self-expression model, called single-cell multimodal self-expressive integration (scMSI), which integrates the heterogeneous multimodal data into a unified manifold space. Specifically, scMSI first learns each omics-specific latent representation and self-expression relationship to consider the characteristics of different omics data by deep self-expressive generative model. Then, scMSI combines these omics-specific self-expression relations through contrastive learning. In such a way, scMSI provides a paradigm to integrate multiple omics data even with weak relation, which effectively achieves the representation learning and data integration into a unified framework. We demonstrate that scMSI provides a cohesive solution for a variety of analysis tasks, such as integration analysis, data denoising, batch correction and spatial domain detection. We have applied scMSI on various single-cell and spatial multimodal datasets to validate its high effectiveness and robustness in diverse data types and application scenarios.

Spatially aware self-representation learning for tissue structure characterization and spatial functional genes identification.

Spatially resolved transcriptomics (SRT) enable the comprehensive characterization of transcriptomic profiles in the context of tissue microenvironments. Unveiling spatial transcriptional heterogeneity needs to effectively incorporate spatial information accounting for the substantial spatial correlation of expression measurements. Here, we develop a computational method, SpaSRL (spatially aware self-representation learning), which flexibly enhances and decodes spatial transcriptional signals to simultaneously achieve spatial domain detection and spatial functional genes identification. This novel tunable spatially aware strategy of SpaSRL not only balances spatial and transcriptional coherence for the two tasks, but also can transfer spatial correlation constraint between them based on a unified model. In addition, this joint analysis by SpaSRL deciphers accurate and fine-grained tissue structures and ensures the effective extraction of biologically informative genes underlying spatial architecture. We verified the superiority of SpaSRL on spatial domain detection, spatial functional genes identification and data denoising using multiple SRT datasets obtained by different platforms and tissue sections. Our results illustrate SpaSRL's utility in flexible integration of spatial information and novel discovery of biological insights from spatial transcriptomic datasets.

Dynamic network biomarker factors orchestrate cell-fate determination at tipping points during hESC differentiation.

The generation of ectoderm, mesoderm, and endoderm layers is the most critical biological process during the gastrulation of embryo development. Such a differentiation process in human embryonic stem cells (hESCs) is an inherently nonlinear multi-stage dynamical process which contain multiple tipping points playing crucial roles in the cell-fate decision. However, the tipping points of the process are largely unknown, letting alone the understanding of the molecular regulation on these critical events. Here by designing a module-based dynamic network biomarker (M-DNB) model, we quantitatively pinpointed two tipping points of the differentiation of hESCs toward definitive endoderm, which leads to the identification of M-DNB factors (FOS, HSF1, MYCN, TP53, and MYC) of this process. We demonstrate that before the tipping points, M-DNB factors are able to maintain the cell states and orchestrate cell-fate determination during hESC (ES)-to-ME and ME-to-DE differentiation processes, which not only leads to better understanding of endodermal specification of hESCs but also reveals the power of the M-DNB model to identify critical transition points with their key factors in diverse biological processes, including cell differentiation and transdifferentiation dynamics.

Lipid peroxidation of immune cells in cancer.

Growing evidence indicates that cellular metabolism is a critical determinant of immune cell viability and function in antitumor immunity and lipid metabolism is important for immune cell activation and adaptation to the tumor microenvironment (TME). Lipid peroxidation is a process in which oxidants attack lipid-containing carbon-carbon double bonds and is an important part of lipid metabolism. In the past decades, studies have shown that lipid peroxidation participates in signal transduction to control cell proliferation, differentiation, and cell death, which is essential for cell function execution and human health. More importantly, recent studies have shown that lipid peroxidation affects immune cell function to modulate tumor immunity and antitumor ability. In this review, we briefly overview the effect of lipid peroxidation on the adaptive and innate immune cell activation and function in TME and discuss the effectiveness and sensitivity of the antitumor ability of immune cells by regulating lipid peroxidation.

A self-contained acoustofluidic platform for biomarker detection.

Self-contained microfluidic platforms with on-chip integration of flow control units, microreactors, (bio)sensors, etc. are ideal systems for point-of-care (POC) testing. However, current approaches such as micropumps and microvalves, increase the cost and the control system, and it is rather difficult to integrate into a single chip. Herein, we demonstrated a versatile acoustofluidic platform actuated by a Lamb wave resonator (LWR) array, in which pumping, mixing, fluidic switching, and particle trapping are all achieved on a single chip. The high-speed microscale acoustic streaming triggered by the LWR in the confined microchannel can be utilized to realize a flow resistor and switch. Variable unidirectional pumping was realized by regulating the relative position of the LWR in various custom-designed microfluidic structures and adoption of different geometric parameters for the microchannel. In addition, to realize quantitative biomarker detection, the on-chip flow resistor, micropump, micromixer and particle trapper were also integrated with a CMOS photo sensor and electronic driver circuit, resulting in an automated handheld microfluidic system with no moving parts. Finally, the acoustofluidic platform was tested for prostate-specific antigen (PSA) sensing, which demonstrates the biocompatibility and applied potency of this proposed self-contained system in POC biomedical applications.

Micropeptide PACMP inhibition elicits synthetic lethal effects by decreasing CtIP and poly(ADP-ribosyl)ation.

Synthetic lethality through combinatorial targeting DNA damage response (DDR) pathways provides exciting anticancer therapeutic benefit. Currently, the long noncoding RNAs (lncRNAs) have been implicated in tumor drug resistance; however, their potential significance in DDR is still largely unknown. Here, we report that a human lncRNA, CTD-2256P15.2, encodes a micropeptide, named PAR-amplifying and CtIP-maintaining micropeptide (PACMP), with a dual function to maintain CtIP abundance and promote poly(ADP-ribosyl)ation. PACMP not only prevents CtIP from ubiquitination through inhibiting the CtIP-KLHL15 association but also directly binds DNA damage-induced poly(ADP-ribose) chains to enhance PARP1-dependent poly(ADP-ribosyl)ation. Targeting PACMP alone inhibits tumor growth by causing a synthetic lethal interaction between CtIP and PARP inhibitions and confers sensitivity to PARP/ATR/CDK4/6 inhibitors, ionizing radiation, epirubicin, and camptothecin. Our findings reveal that a lncRNA-derived micropeptide regulates cancer progression and drug resistance by modulating DDR, whose inhibition could be employed to augment the existing anticancer therapeutic strategies.

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.

scDA: Single cell discriminant analysis for single-cell RNA sequencing data.

Single-cell RNA-sequencing (scRNA-seq) techniques provide unprecedented opportunities to investigate phenotypic and molecular heterogeneity in complex biological systems. However, profiling massive amounts of cells brings great computational challenges to accurately and efficiently characterize diverse cell populations. Single cell discriminant analysis (scDA) solves this problem by simultaneously identifying cell groups and discriminant metagenes based on the construction of cell-by-cell representation graph, and then using them to annotate unlabeled cells in data. We demonstrate scDA is effective to determine cell types, revealing the overall variabilities between cells from eleven data sets. scDA also outperforms several state-of-the-art methods when inferring the labels of new samples. In particular, we found scDA less sensitive to drop-out events and capable to label a mass of cells within or across datasets after learning even from a small set of data. The scDA approach offers a new way to efficiently analyze scRNA-seq profiles of large size or from different batches. scDA was implemented and freely available at https://github.com/ZCCQQWork/scDA.

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

Significantly increasing crop yield is a major and worldwide challenge for food supply and security. It is well-known that rice cultivated at Taoyuan in Yunnan of China can produce the highest yield worldwide. Yet, the gene regulatory mechanism underpinning this ultrahigh yield has been a mystery. Here, we systematically collected the transcriptome data for seven key tissues at different developmental stages using rice cultivated both at Taoyuan as the case group and at another regular rice planting place Jinghong as the control group. We identified the top 24 candidate high-yield genes with their network modules from these well-designed datasets by developing a novel computational systems biology method, i.e., dynamic cross-tissue (DCT) network analysis. We used one of the candidate genes, OsSPL4, whose function was previously unknown, for gene editing experimental validation of the high yield, and confirmed that OsSPL4 significantly affects panicle branching and increases the rice yield. This study, which included extensive field phenotyping, cross-tissue systems biology analyses, and functional validation, uncovered the key genes and gene regulatory networks underpinning the ultrahigh yield of rice. The DCT method could be applied to other plant or animal systems if different phenotypes under various environments with the common genome sequences of the examined sample. DCT can be downloaded from https://github.com/ztpub/DCT.

Multi-view Subspace Clustering Analysis for Aggregating Multiple Heterogeneous Omics Data.

Integration of distinct biological data types could provide a comprehensive view of biological processes or complex diseases. The combinations of molecules responsible for different phenotypes form multiple embedded (expression) subspaces, thus identifying the intrinsic data structure is challenging by regular integration methods. In this paper, we propose a novel framework of "Multi-view Subspace Clustering Analysis (MSCA)," which could measure the local similarities of samples in the same subspace and obtain the global consensus sample patterns (structures) for multiple data types, thereby comprehensively capturing the underlying heterogeneity of samples. Applied to various synthetic datasets, MSCA performs effectively to recognize the predefined sample patterns, and is robust to data noises. Given a real biological dataset, i.e., Cancer Cell Line Encyclopedia (CCLE) data, MSCA successfully identifies cell clusters of common aberrations across cancer types. A remarkable superiority over the state-of-the-art methods, such as iClusterPlus, SNF, and ANF, has also been demonstrated in our simulation and case studies.

Broadband terahertz antireflective microstructures on quartz crystal surface by CO2 laser micro-processing.

Anti-reflection (AR) coating is a critical technology and an ongoing challenge for terahertz systems. The subwavelength structure (SWS) is an effective AR method, whereas the current manufacturing techniques, such as chemical etching and ultrafast laser processing, are low-efficient and low-quality for processing structures at the hundred-micron scale on hard brittle materials. We present a study of broadband SWSs directly ablated on the surface of quartz crystal by precisely controlled CO2 laser pulses, instead of commonly used ultra-fast lasers. The processing time of SWS can be shortened by two orders of magnitude compared with that by ultra-fast laser pulses. The SWS samples exhibit excellent AR properties with maximum transmittance of 97% at 0.71 THz, peak transmittance improvement of 13.5%, and optimal efficiency spectrum of 0.28-1.21 THz with transmittance >90%. The AR properties of SWS samples are in agreement with the simulated expectation and exist over a wide range of incidence angles up to ∼40°. The imaging of an object using SWS as the substrate shows an obvious improvement in imaging quality. We present an efficient and practical way to improve the transmission of optical components of materials, such as quartz crystal, alumina, and sapphire, in the terahertz band.

Hydrofluoric acid-based etching effect on surface pit, crack, and scratch and laser damage site of fused silica optics.

In a large-scale, high-power laser facility, fused silica optics plays an irreplaceable role to transmit extremely intense lasers. However, the surface fractures, such as surface pit, crack, and scratch and laser damage site, of fused silica optics will shorten the lifetime of the optics and thus limit the output performance of the laser facility. In this work, besides experimental study, finite difference time domain (FDTD) simulation is performed to study hydrofluoric acid-based (HF-based) etching effect on the surface fractures. The effect of local surface curvature on etching rate is discussed and an explicit local-curvature-dependent etching model is proposed. Based on this model, the result from FDTD simulation qualitatively agrees very well with that of the experiment. It is demonstrated that the FDTD simulation is efficient to predict the morphological evolution of the surface fractures during etching. In addition, it is found that the surface fractures will be passivated and HF-based etching can greatly suppress the laser-damage growth of laser-induced damage to the surface site of fused silica optics.

Effect of pressure gradient and new phases for 1,3,5-trinitrohexahydro-s-triazine (RDX) under high pressures.

Herein, pressure-induced phase transitions of RDX up to 50 GPa were systematically studied under different compression conditions. Precise phase transition points were obtained based on high-quality Raman spectra with small pressure intervals. This favors the correctness of the theoretical formula for detonation and the design of a precision weapon. The experimental results indicated that α-RDX immediately transformed to γ-RDX at 3.5 GPa due to hydrostatic conditions and possible interaction between the penetrating helium and RDX, with helium gas as the pressure-transmitting medium (PTM). Mapping of pressure distribution in samples demonstrates that the pressure gradient is generated in the chamber and independent of other PTMs. The gradient induced the first phase transition starts at 2.3 GPa and completed at 4.1 GPa. The larger pressure gradient promoted phase transition in advance under higher pressures. Experimental results supported that there existed two conformers of AAI and AAE for γ-RDX, as proposed by another group. δ-RDX was considered to only occur in a hydrostatic environment around 18 GPa using helium as the PTM. This study confirms that δ-RDX is independent of PTM and exists under non-hydrostatic conditions. Evidence for a new phase (ζ) was found at about 28 GPa. These 4 phases have also been verified via XRD under high pressures. In addition to this, another new phase (η) may exist above 38 GPa, and it needs to be further confirmed in the future. Moreover, all the phase transitions were reversible after the pressure was released, and original α-RDX was always obtained at ambient pressure.

TMCO1 is essential for ovarian follicle development by regulating ER Ca2+ store of granulosa cells.

TMCO1 (transmembrane and coiled-coil domains 1) is an endoplasmic reticulum (ER) transmembrane protein that actively prevents Ca2+ stores from overfilling. To characterize its physiological function(s), we generated Tmco1-/- knockout (KO) mice. In addition to the main clinical features of human cerebrofaciothoracic (CFT) dysplasia spectrum, Tmco1-/- females manifest gradual loss of ovarian follicles, impaired ovarian follicle development, and subfertility with a phenotype analogous to the premature ovarian failure (POF) in women. In line with the role of TMCO1 as a Ca2+ load-activated Ca2+ channel, we have detected a supernormal Ca2+ signaling in Tmco1-/- granulosa cells (GCs). Interestingly, although spontaneous Ca2+ oscillation pattern was altered, ER Ca2+ stores of germinal vesicle (GV) stage oocytes and metaphase II (MII) arrested eggs were normal upon Tmco1 ablation. Combined with RNA-sequencing analysis, we also detected increased ER stress-mediated apoptosis and enhanced reactive oxygen species (ROS) level in Tmco1-/- GCs, indicating the dysfunctions of GCs upon TMCO1 deficiency. Taken together, these results reveal that TMCO1 is essential for ovarian follicle development and female fertility by maintaining ER Ca2+ homeostasis of GCs, disruption of which causes ER stress-mediated apoptosis and increased cellular ROS level in GCs and thus leads to impaired ovarian follicle development.

Local network component analysis for quantifying transcription factor activities.

Transcription factors (TFs) could regulate physiological transitions or determine stable phenotypic diversity. The accurate estimation on TF regulatory signals or functional activities is of great significance to guide biological experiments or elucidate molecular mechanisms, but still remains challenging. Traditional methods identify TF regulatory signals at the population level, which masks heterogeneous regulation mechanisms in individuals or subgroups, thus resulting in inaccurate analyses. Here, we propose a novel computational framework, namely local network component analysis (LNCA), to exploit data heterogeneity and automatically quantify accurate transcription factor activity (TFA) in practical terms, through integrating the partitioned expression sets (i.e., local information) and prior TF-gene regulatory knowledge. Specifically, LNCA adopts an adaptive optimization strategy, which evaluates the local similarities of regulation controls and corrects biases during data integration, to construct the TFA landscape. In particular, we first numerically demonstrate the effectiveness of LNCA for the simulated data sets, compared with traditional methods, such as FastNCA, ROBNCA and NINCA. Then, we apply our model to two real data sets with implicit temporal or spatial regulation variations. The results show that LNCA not only recognizes the periodic mode along the S. cerevisiae cell cycle process, but also substantially outperforms over other methods in terms of accuracy and consistency. In addition, the cross-validation study for glioblastomas multiforme (GBM) indicates that the TFAs, identified by LNCA, can better distinguish clinically distinct tumor groups than the expression values of the corresponding TFs, thus opening a new way to classify tumor subtypes and also providing a novel insight into cancer heterogeneity. AVAILABILITY: LNCA was implemented as a Matlab package, which is available at http://sysbio.sibcb.ac.cn/cb/chenlab/software.htm/LNCApackage_0.1.rar.

Pattern fusion analysis by adaptive alignment of multiple heterogeneous omics data.

MOTIVATION: Integrating different omics profiles is a challenging task, which provides a comprehensive way to understand complex diseases in a multi-view manner. One key for such an integration is to extract intrinsic patterns in concordance with data structures, so as to discover consistent information across various data types even with noise pollution. Thus, we proposed a novel framework called 'pattern fusion analysis' (PFA), which performs automated information alignment and bias correction, to fuse local sample-patterns (e.g. from each data type) into a global sample-pattern corresponding to phenotypes (e.g. across most data types). In particular, PFA can identify significant sample-patterns from different omics profiles by optimally adjusting the effects of each data type to the patterns, thereby alleviating the problems to process different platforms and different reliability levels of heterogeneous data. RESULTS: To validate the effectiveness of our method, we first tested PFA on various synthetic datasets, and found that PFA can not only capture the intrinsic sample clustering structures from the multi-omics data in contrast to the state-of-the-art methods, such as iClusterPlus, SNF and moCluster, but also provide an automatic weight-scheme to measure the corresponding contributions by data types or even samples. In addition, the computational results show that PFA can reveal shared and complementary sample-patterns across data types with distinct signal-to-noise ratios in Cancer Cell Line Encyclopedia (CCLE) datasets, and outperforms over other works at identifying clinically distinct cancer subtypes in The Cancer Genome Atlas (TCGA) datasets. AVAILABILITY AND IMPLEMENTATION: PFA has been implemented as a Matlab package, which is available at http://www.sysbio.ac.cn/cb/chenlab/images/PFApackage_0.1.rar . CONTACT: lnchen@sibs.ac.cn , liujuan@whu.edu.cn or zengtao@sibs.ac.cn. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Comparative network stratification analysis for identifying functional interpretable network biomarkers.

BACKGROUND: A major challenge of bioinformatics in the era of precision medicine is to identify the molecular biomarkers for complex diseases. It is a general expectation that these biomarkers or signatures have not only strong discrimination ability, but also readable interpretations in a biological sense. Generally, the conventional expression-based or network-based methods mainly capture differential genes or differential networks as biomarkers, however, such biomarkers only focus on phenotypic discrimination and usually have less biological or functional interpretation. Meanwhile, the conventional function-based methods could consider the biomarkers corresponding to certain biological functions or pathways, but ignore the differential information of genes, i.e., disregard the active degree of particular genes involved in particular functions, thereby resulting in less discriminative ability on phenotypes. Hence, it is strongly demanded to develop elaborate computational methods to directly identify functional network biomarkers with both discriminative power on disease states and readable interpretation on biological functions. RESULTS: In this paper, we present a new computational framework based on an integer programming model, named as Comparative Network Stratification (CNS), to extract functional or interpretable network biomarkers, which are of strongly discriminative power on disease states and also readable interpretation on biological functions. In addition, CNS can not only recognize the pathogen biological functions disregarded by traditional Expression-based/Network-based methods, but also uncover the active network-structures underlying such dysregulated functions underestimated by traditional Function-based methods. To validate the effectiveness, we have compared CNS with five state-of-the-art methods, i.e. GSVA, Pathifier, stSVM, frSVM and AEP on four datasets of different complex diseases. The results show that CNS can enhance the discriminative power of network biomarkers, and further provide biologically interpretable information or disease pathogenic mechanism of these biomarkers. A case study on type 1 diabetes (T1D) demonstrates that CNS can identify many dysfunctional genes and networks previously disregarded by conventional approaches. CONCLUSION: Therefore, CNS is actually a powerful bioinformatics tool, which can identify functional or interpretable network biomarkers with both discriminative power on disease states and readable interpretation on biological functions. CNS was implemented as a Matlab package, which is available at http://www.sysbio.ac.cn/cb/chenlab/images/CNSpackage_0.1.rar .