Identification of disease modules using higher-order network structure.

Motivation: Higher-order interaction patterns among proteins have the potential to reveal mechanisms behind molecular processes and diseases. While clustering methods are used to identify functional groups within molecular interaction networks, these methods largely focus on edge density and do not explicitly take into consideration higher-order interactions. Disease genes in these networks have been shown to exhibit rich higher-order structure in their vicinity, and considering these higher-order interaction patterns in network clustering have the potential to reveal new disease-associated modules. Results: We propose a higher-order community detection method which identifies community structure in networks with respect to specific higher-order connectivity patterns beyond edges. Higher-order community detection on four different protein-protein interaction networks identifies biologically significant modules and disease modules that conventional edge-based clustering methods fail to discover. Higher-order clusters also identify disease modules from genome-wide association study data, including new modules that were not discovered by top-performing approaches in a Disease Module DREAM Challenge. Our approach provides a more comprehensive view of community structure that enables us to predict new disease-gene associations. Availability and implementation: https://github.com/Reed-CompBio/graphlet-clustering.

From network analysis to experimental validation: identification of regulators of non-muscle myosin II contractility using the folded-gastrulation signaling pathway.

The morphogenetic process of apical constriction, which relies on non-muscle myosin II (NMII) generated constriction of apical domains of epithelial cells, is key to the development of complex cellular patterns. Apical constriction occurs in almost all multicellular organisms, but one of the most well-characterized systems is the Folded-gastrulation (Fog)-induced apical constriction that occurs in Drosophila. The binding of Fog to its cognizant receptors Mist/Smog results in a signaling cascade that leads to the activation of NMII-generated contractility. Despite our knowledge of key molecular players involved in Fog signaling, we sought to explore whether other proteins have an undiscovered role in its regulation. We developed a computational method to predict unidentified candidate NMII regulators using a network of pairwise protein-protein interactions called an interactome. We first constructed a Drosophila interactome of over 500,000 protein-protein interactions from several databases that curate high-throughput experiments. Next, we implemented several graph-based algorithms that predicted 14 proteins potentially involved in Fog signaling. To test these candidates, we used RNAi depletion in combination with a cellular contractility assay in Drosophila S2R + cells, which respond to Fog by contracting in a stereotypical manner. Of the candidates we screened using this assay, two proteins, the serine/threonine phosphatase Flapwing and the putative guanylate kinase CG11811 were demonstrated to inhibit cellular contractility when depleted, suggestive of their roles as novel regulators of the Fog pathway.

Growing Directed Acyclic Graphs: Optimization Functions for Pathway Reconstruction Algorithms.

A major challenge in molecular systems biology is to understand how proteins work to transmit external signals to changes in gene expression. Computationally reconstructing these signaling pathways from protein interaction networks can help understand what is missing from existing pathway databases. We formulate a new pathway reconstruction problem, one that iteratively grows directed acyclic graphs (DAGs) from a set of starting proteins in a protein interaction network. We present an algorithm that provably returns the optimal DAGs for two different cost functions and evaluate the pathway reconstructions when applied to six diverse signaling pathways from the NetPath database. The optimal DAGs outperform an existing k-shortest paths method for pathway reconstruction, and the new reconstructions are enriched for different biological processes. Growing DAGs is a promising step toward reconstructing pathways that provably optimize a specific cost function.

Transformer Neural Networks for Protein Family and Interaction Prediction Tasks.

The scientific community is rapidly generating protein sequence information, but only a fraction of these proteins can be experimentally characterized. While promising deep learning approaches for protein prediction tasks have emerged, they have computational limitations or are designed to solve a specific task. We present a Transformer neural network that pre-trains task-agnostic sequence representations. This model is fine-tuned to solve two different protein prediction tasks: protein family classification and protein interaction prediction. Our method is comparable to existing state-of-the-art approaches for protein family classification while being much more general than other architectures. Further, our method outperforms other approaches for protein interaction prediction for two out of three different scenarios that we generated. These results offer a promising framework for fine-tuning the pre-trained sequence representations for other protein prediction tasks.

Reconciling Signaling Pathway Databases with Network Topologies.

A major goal of molecular systems biology is to understand the coordinated function of genes or proteins in response to cellular signals and to understand these dynamics in the context of disease. Signaling pathway databases such as KEGG, NetPath, NCI-PID, and Panther describe the molecular interactions involved in different cellular responses. While the same pathway may be present in different databases, prior work has shown that the particular proteins and interactions differ across database annotations. However, to our knowledge no one has attempted to quantify their structural differences. It is important to characterize artifacts or other biases within pathway databases, which can provide a more informed interpretation for downstream analyses. In this work we consider signaling pathways as graphs and we use topological measures to study their structure. We find that topological characterization using graphlets (small, connected subgraphs) distinguishes signaling pathways from appropriate null models of interaction networks. Next, we quantify topological similarity across pathway databases. Our analysis reveals that the pathways harbor database-specific characteristics implying that even though these databases describe the same pathways, they tend to be systematically different from one another. We show that pathway-specific topology can be uncovered after accounting for database-specific structure. This work presents the first step towards elucidating common pathway structure beyond their specific database annotations.Data Availability: https://github.com/Reed-CompBio/pathway-reconciliation.

Cabozantinib and dasatinib synergize to induce tumor regression in non-clear cell renal cell carcinoma.

The lack of effective treatment options for advanced non-clear cell renal cell carcinoma (NCCRCC) is a critical unmet clinical need. Applying a high-throughput drug screen to multiple human kidney cancer cells, we identify the combination of the VEGFR-MET inhibitor cabozantinib and the SRC inhibitor dasatinib acts synergistically in cells to markedly reduce cell viability. Importantly, the combination is well tolerated and causes tumor regression in vivo. Transcriptional and phosphoproteomic profiling reveals that the combination converges to downregulate the MAPK-ERK signaling pathway, a result not predicted by single-agent analysis alone. Correspondingly, the addition of a MEK inhibitor synergizes with either dasatinib or cabozantinib to increase its efficacy. This study, by using approved, clinically relevant drugs, provides the rationale for the design of effective combination treatments in NCCRCC that can be rapidly translated to the clinic.

Graphery: interactive tutorials for biological network algorithms.

Networks have been an excellent framework for modeling complex biological information, but the methodological details of network-based tools are often described for a technical audience. We have developed Graphery, an interactive tutorial webserver that illustrates foundational graph concepts frequently used in network-based methods. Each tutorial describes a graph concept along with executable Python code that can be interactively run on a graph. Users navigate each tutorial using their choice of real-world biological networks that highlight the diverse applications of network algorithms. Graphery also allows users to modify the code within each tutorial or write new programs, which all can be executed without requiring an account. Graphery accepts ideas for new tutorials and datasets that will be shaped by both computational and biological researchers, growing into a community-contributed learning platform. Graphery is available at https://graphery.reedcompbio.org/.

The Drosophila melanogaster Rab GAP RN-tre cross-talks with the Rho1 signaling pathway to regulate nonmuscle myosin II localization and function.

To identify novel regulators of nonmuscle myosin II (NMII) we performed an image-based RNA interference screen using stable Drosophila melanogaster S2 cells expressing the enhanced green fluorescent protein (EGFP)-tagged regulatory light chain (RLC) of NMII and mCherry-Actin. We identified the Rab-specific GTPase-activating protein (GAP) RN-tre as necessary for the assembly of NMII RLC into contractile actin networks. Depletion of RN-tre led to a punctate NMII phenotype, similar to what is observed following depletion of proteins in the Rho1 pathway. Depletion of RN-tre also led to a decrease in active Rho1 and a decrease in phosphomyosin-positive cells by immunostaining, while expression of constitutively active Rho or Rho-kinase (Rok) rescues the punctate phenotype. Functionally, RN-tre depletion led to an increase in actin retrograde flow rate and cellular contractility in S2 and S2R+ cells, respectively. Regulation of NMII by RN-tre is only partially dependent on its GAP activity as overexpression of constitutively active Rabs inactivated by RN-tre failed to alter NMII RLC localization, while a GAP-dead version of RN-tre partially restored phosphomyosin staining. Collectively, our results suggest that RN-tre plays an important regulatory role in NMII RLC distribution, phosphorylation, and function, likely through Rho1 signaling and putatively serving as a link between the secretion machinery and actomyosin contractility.

MsPAC: a tool for haplotype-phased structural variant detection.

SUMMARY: While next-generation sequencing (NGS) has dramatically increased the availability of genomic data, phased genome assembly and structural variant (SV) analyses are limited by NGS read lengths. Long-read sequencing from Pacific Biosciences and NGS barcoding from 10x Genomics hold the potential for far more comprehensive views of individual genomes. Here, we present MsPAC, a tool that combines both technologies to partition reads, assemble haplotypes (via existing software) and convert assemblies into high-quality, phased SV predictions. MsPAC represents a framework for haplotype-resolved SV calls that moves one step closer to fully resolved, diploid genomes. AVAILABILITY AND IMPLEMENTATION: https://github.com/oscarlr/MsPAC. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Distance measures for tumor evolutionary trees.

MOTIVATION: There has been recent increased interest in using algorithmic methods to infer the evolutionary tree underlying the developmental history of a tumor. Quantitative measures that compare such trees are vital to a number of different applications including benchmarking tree inference methods and evaluating common inheritance patterns across patients. However, few appropriate distance measures exist, and those that do have low resolution for differentiating trees or do not fully account for the complex relationship between tree topology and the inheritance of the mutations labeling that topology. RESULTS: Here, we present two novel distance measures, Common Ancestor Set distance (CASet) and Distinctly Inherited Set Comparison distance (DISC), that are specifically designed to account for the subclonal mutation inheritance patterns characteristic of tumor evolutionary trees. We apply CASet and DISC to multiple simulated datasets and two breast cancer datasets and show that our distance measures allow for more nuanced and accurate delineation between tumor evolutionary trees than existing distance measures. AVAILABILITY AND IMPLEMENTATION: Implementations of CASet and DISC are freely available at: https://bitbucket.org/oesperlab/stereodist. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Integrating protein localization with automated signaling pathway reconstruction.

BACKGROUND: Understanding cellular responses via signal transduction is a core focus in systems biology. Tools to automatically reconstruct signaling pathways from protein-protein interactions (PPIs) can help biologists generate testable hypotheses about signaling. However, automatic reconstruction of signaling pathways suffers from many interactions with the same confidence score leading to many equally good candidates. Further, some reconstructions are biologically misleading due to ignoring protein localization information. RESULTS: We propose LocPL, a method to improve the automatic reconstruction of signaling pathways from PPIs by incorporating information about protein localization in the reconstructions. The method relies on a dynamic program to ensure that the proteins in a reconstruction are localized in cellular compartments that are consistent with signal transduction from the membrane to the nucleus. LocPL and existing reconstruction algorithms are applied to two PPI networks and assessed using both global and local definitions of accuracy. LocPL produces more accurate and biologically meaningful reconstructions on a versatile set of signaling pathways. CONCLUSION: LocPL is a powerful tool to automatically reconstruct signaling pathways from PPIs that leverages cellular localization information about proteins. The underlying dynamic program and signaling model are flexible enough to study cellular signaling under different settings of signaling flow across the cellular compartments.

Hypergraph-based connectivity measures for signaling pathway topologies.

Characterizing cellular responses to different extrinsic signals is an active area of research, and curated pathway databases describe these complex signaling reactions. Here, we revisit a fundamental question in signaling pathway analysis: are two molecules "connected" in a network? This question is the first step towards understanding the potential influence of molecules in a pathway, and the answer depends on the choice of modeling framework. We examined the connectivity of Reactome signaling pathways using four different pathway representations. We find that Reactome is very well connected as a graph, moderately well connected as a compound graph or bipartite graph, and poorly connected as a hypergraph (which captures many-to-many relationships in reaction networks). We present a novel relaxation of hypergraph connectivity that iteratively increases connectivity from a node while preserving the hypergraph topology. This measure, B-relaxation distance, provides a parameterized transition between hypergraph connectivity and graph connectivity. B-relaxation distance is sensitive to the presence of small molecules that participate in many functionally unrelated reactions in the network. We also define a score that quantifies one pathway's downstream influence on another, which can be calculated as B-relaxation distance gradually relaxes the connectivity constraint in hypergraphs. Computing this score across all pairs of 34 Reactome pathways reveals pairs of pathways with statistically significant influence. We present two such case studies, and we describe the specific reactions that contribute to the large influence score. Finally, we investigate the ability for connectivity measures to capture functional relationships among proteins, and use the evidence channels in the STRING database as a benchmark dataset. STRING interactions whose proteins are B-connected in Reactome have statistically significantly higher scores than interactions connected in the bipartite graph representation. Our method lays the groundwork for other generalizations of graph-theoretic concepts to hypergraphs in order to facilitate signaling pathway analysis.

Invasive ductal carcinoma detected within a fibroadenolipoma through digital breast tomosynthesis.

A 52-year-old patient referred to our hospital for a screening mammogram showed a suspicious new architectural distortion. Previously, a fibroadenolipoma within the right breast was diagnosed clinically and radiologically. Further work-up with tomosynthesis, magnetic resonance imaging, and magnetic resonance-guided biopsy showed an invasive ductal carcinoma within the fibroadenolipoma, which are usually benign breast lesions not associated with malignancy. This case report offers a review of the literature and a discussion of signs, which should alert the radiologist.

Network-based prediction of polygenic disease genes involved in cell motility.

BACKGROUND: Schizophrenia and autism are examples of polygenic diseases caused by a multitude of genetic variants, many of which are still poorly understood. Recently, both diseases have been associated with disrupted neuron motility and migration patterns, suggesting that aberrant cell motility is a phenotype for these neurological diseases. RESULTS: We formulate the POLYGENIC DISEASE PHENOTYPE Problem which seeks to identify candidate disease genes that may be associated with a phenotype such as cell motility. We present a machine learning approach to solve this problem for schizophrenia and autism genes within a brain-specific functional interaction network. Our method outperforms peer semi-supervised learning approaches, achieving better cross-validation accuracy across different sets of gold-standard positives. We identify top candidates for both schizophrenia and autism, and select six genes labeled as schizophrenia positives that are predicted to be associated with cell motility for follow-up experiments. CONCLUSIONS: Candidate genes predicted by our method suggest testable hypotheses about these genes’ role in cell motility regulation, offering a framework for generating predictions for experimental validation.

A Cell-based Assay to Investigate Non-muscle Myosin II Contractility via the Folded-gastrulation Signaling Pathway in Drosophila S2R+ Cells.

We have developed a cell-based assay using Drosophila cells that recapitulates apical constriction initiated by folded gastrulation (Fog), a secreted epithelial morphogen. In this assay, Fog is used as an agonist to activate Rho through a signaling cascade that includes a G-protein-coupled receptor (Mist), a Gα12/13 protein (Concertina/Cta), and a PDZ-domain-containing guanine nucleotide exchange factor (RhoGEF2). Fog signaling results in the rapid and dramatic reorganization of the actin cytoskeleton to form a contractile purse string. Soluble Fog is collected from a stable cell line and applied ectopically to S2R+ cells, leading to morphological changes like apical constriction, a process observed during developmental processes such as gastrulation. This assay is amenable to high-throughput screening and, using RNAi, can facilitate the identification of additional genes involved in this pathway.

Metabolic reprogramming ensures cancer cell survival despite oncogenic signaling blockade.

There is limited knowledge about the metabolic reprogramming induced by cancer therapies and how this contributes to therapeutic resistance. Here we show that although inhibition of PI3K-AKT-mTOR signaling markedly decreased glycolysis and restrained tumor growth, these signaling and metabolic restrictions triggered autophagy, which supplied the metabolites required for the maintenance of mitochondrial respiration and redox homeostasis. Specifically, we found that survival of cancer cells was critically dependent on phospholipase A2 (PLA2) to mobilize lysophospholipids and free fatty acids to sustain fatty acid oxidation and oxidative phosphorylation. Consistent with this, we observed significantly increased lipid droplets, with subsequent mobilization to mitochondria. These changes were abrogated in cells deficient for the essential autophagy gene ATG5 Accordingly, inhibition of PLA2 significantly decreased lipid droplets, decreased oxidative phosphorylation, and increased apoptosis. Together, these results describe how treatment-induced autophagy provides nutrients for cancer cell survival and identifies novel cotreatment strategies to override this survival advantage.

Pathway Analysis with Signaling Hypergraphs.

Signaling pathways play an important role in the cell's response to its environment. Signaling pathways are often represented as directed graphs, which are not adequate for modeling reactions such as complex assembly and dissociation, combinatorial regulation, and protein activation/inactivation. More accurate representations such as directed hypergraphs remain underutilized. In this paper, we present an extension of a directed hypergraph that we call a signaling hypergraph. We formulate a problem that asks what proteins and interactions must be involved in order to stimulate a specific response downstream of a signaling pathway. We relate this problem to computing the shortest acyclic B-hyperpath in a signaling hypergraph-an NP-hard problem-and present a mixed integer linear program to solve it. We demonstrate that the shortest hyperpaths computed in signaling hypergraphs are far more informative than shortest paths, Steiner trees, and subnetworks containing many short paths found in corresponding graph representations. Our results illustrate the potential of signaling hypergraphs as an improved representation of signaling pathways and motivate the development of novel hypergraph algorithms.

GraphSpace: stimulating interdisciplinary collaborations in network biology.

SUMMARY: Networks have become ubiquitous in systems biology. Visualization is a crucial component in their analysis. However, collaborations within research teams in network biology are hampered by software systems that are either specific to a computational algorithm, create visualizations that are not biologically meaningful, or have limited features for sharing networks and visualizations. We present GraphSpace, a web-based platform that fosters team science by allowing collaborating research groups to easily store, interact with, layout and share networks. AVAILABILITY AND IMPLEMENTATION: Anyone can upload and share networks at http://graphspace.org. In addition, the GraphSpace code is available at http://github.com/Murali-group/graphspace if a user wants to run his or her own server. CONTACT: murali@cs.vt.edu. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.