How Machine Learning Will Transform Biomedicine.

This Perspective explores the application of machine learning toward improved diagnosis and treatment. We outline a vision for how machine learning can transform three broad areas of biomedicine: clinical diagnostics, precision treatments, and health monitoring, where the goal is to maintain health through a range of diseases and the normal aging process. For each area, early instances of successful machine learning applications are discussed, as well as opportunities and challenges for machine learning. When these challenges are met, machine learning promises a future of rigorous, outcomes-based medicine with detection, diagnosis, and treatment strategies that are continuously adapted to individual and environmental differences.

The Human Tumor Atlas Network: Charting Tumor Transitions across Space and Time at Single-Cell Resolution.

Crucial transitions in cancer-including tumor initiation, local expansion, metastasis, and therapeutic resistance-involve complex interactions between cells within the dynamic tumor ecosystem. Transformative single-cell genomics technologies and spatial multiplex in situ methods now provide an opportunity to interrogate this complexity at unprecedented resolution. The Human Tumor Atlas Network (HTAN), part of the National Cancer Institute (NCI) Cancer Moonshot Initiative, will establish a clinical, experimental, computational, and organizational framework to generate informative and accessible three-dimensional atlases of cancer transitions for a diverse set of tumor types. This effort complements both ongoing efforts to map healthy organs and previous large-scale cancer genomics approaches focused on bulk sequencing at a single point in time. Generating single-cell, multiparametric, longitudinal atlases and integrating them with clinical outcomes should help identify novel predictive biomarkers and features as well as therapeutically relevant cell types, cell states, and cellular interactions across transitions. The resulting tumor atlases should have a profound impact on our understanding of cancer biology and have the potential to improve cancer detection, prevention, and therapeutic discovery for better precision-medicine treatments of cancer patients and those at risk for cancer.

Computational multiplex panel reduction to maximize information retention in breast cancer tissue microarrays.

Recent state-of-the-art multiplex imaging techniques have expanded the depth of information that can be captured within a single tissue sample by allowing for panels with dozens of markers. Despite this increase in capacity, space on the panel is still limited due to technical artifacts, tissue loss, and long imaging acquisition time. As such, selecting which markers to include on a panel is important, since removing important markers will result in a loss of biologically relevant information, but identifying redundant markers will provide a room for other markers. To address this, we propose computational approaches to determine the amount of shared information between markers and select an optimally reduced panel that captures maximum amount of information with the fewest markers. Here we examine several panel selection approaches and evaluate them based on their ability to reconstruct the full panel images and information within breast cancer tissue microarray datasets using cyclic immunofluorescence as a proof of concept. We show that all methods perform adequately and can re-capture cell types using only 18 of 25 markers (72% of the original panel size). The correlation-based selection methods achieved the best single-cell marker mean intensity predictions with a Spearman correlation of 0.90 with the reduced panel. Using the proposed methods shown here, it is possible for researchers to design more efficient multiplex imaging panels that maximize the amount of information retained with the limited number of markers with respect to certain evaluation metrics and architecture biases.

Targeted Nanoparticle for Co-delivery of HER2 siRNA and a Taxane to Mirror the Standard Treatment of HER2+ Breast Cancer: Efficacy in Breast Tumor and Brain Metastasis.

The first-line treatment of advanced and metastatic human epidermal growth factor receptor type 2 (HER2+) breast cancer requires two HER2-targeting antibodies (trastuzumab and pertuzumab) and a taxane (docetaxel or paclitaxel). The three-drug regimen costs over $320,000 per treatment course, requires a 4 h infusion time, and has many adverse side effects, while achieving only 18 months of progression-free survival. To replace this regimen, reduce infusion time, and enhance efficacy, a single therapeutic is developed based on trastuzumab-conjugated nanoparticles for co-delivering docetaxel and siRNA against HER2 (siHER2). The optimal nanoconstruct has a hydrodynamic size of 100 nm and specifically treats HER2+ breast cancer cells over organ-derived normal cells. In a drug-resistant orthotopic HER2+ HCC1954 tumor mouse model, the nanoconstruct inhibits tumor growth more effectively than the docetaxel and trastuzumab combination. When coupled with microbubble-assisted focused ultrasound that transiently disrupts the blood brain barrier, the nanoconstruct inhibits the growth of trastuzumab-resistant HER2+ BT474 tumors residing in the brains of mice. The nanoconstruct has a favorable safety profile in cells and in mice. Combination therapies have become the cornerstone of cancer treatment and this versatile nanoparticle platform can co-deliver multiple therapeutic types to ensure that they reach the target cells at the same time to realize their synergy.

Sensitivity to targeted therapy differs between HER2-amplified breast cancer cells harboring kinase and helical domain mutations in PIK3CA.

BACKGROUND: HER2-amplified breast cancer is a clinically defined subtype of breast cancer for which there are multiple viable targeted therapies. Resistance to these targeted therapies is a common problem, but the mechanisms by which resistance occurs remain incompletely defined. One mechanism that has been proposed is through mutation of genes in the PI3-kinase pathway. Intracellular signaling from the HER2 pathway can occur through PI3-kinase, and mutations of the encoding gene PIK3CA are known to be oncogenic. Mutations in PIK3CA co-occur with HER2-amplification in ~ 20% of cases within the HER2-amplified subtype. METHODS: We generated isogenic knockin mutants of each PIK3CA hotspot mutation in HER2-amplified breast cancer cells using adeno-associated virus-mediated gene targeting. Isogenic clones were analyzed using a combinatorial drug screen to determine differential responses to HER2-targeted therapy. Western blot analysis and immunofluorescence uncovered unique intracellular signaling dynamics in cells resistant to HER2-targeted therapy. Subsequent combinatorial drug screens were used to explore neuregulin-1-mediated resistance to HER2-targeted therapy. Finally, results from in vitro experiments were extrapolated to publicly available datasets. RESULTS: Treatment with HER2-targeted therapy reveals that mutations in the kinase domain (H1047R) but not the helical domain (E545K) increase resistance to lapatinib. Mechanistically, sustained AKT signaling drives lapatinib resistance in cells with the kinase domain mutation, as demonstrated by staining for the intracellular product of PI3-kinase, PIP3. This resistance can be overcome by co-treatment with an inhibitor to the downstream kinase AKT. Additionally, knockout of the PIP3 phosphatase, PTEN, phenocopies this result. We also show that neuregulin-1, a ligand for HER-family receptors, confers resistance to cells harboring either hotspot mutation and modulates response to combinatorial therapy. Finally, we show clinical evidence that the hotspot mutations have distinct expression profiles related to therapeutic resistance through analysis of TCGA and METABRIC data cohorts. CONCLUSION: Our results demonstrate unique intracellular signaling differences depending on which mutation in PIK3CA the cell harbors. Only mutations in the kinase domain fully activate the PI3-kinase signaling pathway and maintain downstream signaling in the presence of HER2 inhibition. Moreover, we show there is potentially clinical importance in understanding both the PIK3CA mutational status and levels of neuregulin-1 expression in patients with HER2-amplified breast cancer treated with targeted therapy and that these problems warrant further pre-clinical and clinical testing.

Inferring causal molecular networks: empirical assessment through a community-based effort.

It remains unclear whether causal, rather than merely correlational, relationships in molecular networks can be inferred in complex biological settings. Here we describe the HPN-DREAM network inference challenge, which focused on learning causal influences in signaling networks. We used phosphoprotein data from cancer cell lines as well as in silico data from a nonlinear dynamical model. Using the phosphoprotein data, we scored more than 2,000 networks submitted by challenge participants. The networks spanned 32 biological contexts and were scored in terms of causal validity with respect to unseen interventional data. A number of approaches were effective, and incorporating known biology was generally advantageous. Additional sub-challenges considered time-course prediction and visualization. Our results suggest that learning causal relationships may be feasible in complex settings such as disease states. Furthermore, our scoring approach provides a practical way to empirically assess inferred molecular networks in a causal sense.

Robust Cell Detection and Segmentation for Image Cytometry Reveal Th17 Cell Heterogeneity.

Image cytometry enables quantitative cell characterization with preserved tissue architecture; thus, it has been highlighted in the advancement of multiplex immunohistochemistry (IHC) and digital image analysis in the context of immune-based biomarker monitoring associated with cancer immunotherapy. However, one of the challenges in the current image cytometry methodology is a technical limitation in the segmentation of nuclei and cellular components particularly in heterogeneously stained cancer tissue images. To improve the detection and specificity of single-cell segmentation in hematoxylin-stained images (which can be utilized for recently reported 12-biomarker chromogenic sequential multiplex IHC), we adapted a segmentation algorithm previously developed for hematoxlin and eosin-stained images, where morphological features are extracted based on Gabor-filtering, followed by stacking of image pixels into n-dimensional feature space and unsupervised clustering of individual pixels. Our proposed method showed improved sensitivity and specificity in comparison with standard segmentation methods. Replacing previously proposed methods with our method in multiplex IHC/image cytometry analysis, we observed higher detection of cell lineages including relatively rare TH 17 cells, further enabling sub-population analysis into TH 1-like and TH 2-like phenotypes based on T-bet and GATA3 expression. Interestingly, predominance of TH 2-like TH 17 cells was associated with human papilloma virus (HPV)-negative status of oropharyngeal squamous cell carcinoma of head and neck, known as a poor-prognostic subtype in comparison with HPV-positive status. Furthermore, TH 2-like TH 17 cells in HPV-negative head and neck cancer tissues were spatiotemporally correlated with CD66b+ granulocytes, presumably associated with an immunosuppressive microenvironment. Our cell segmentation method for multiplex IHC/image cytometry potentially contributes to in-depth immune profiling and spatial association, leading to further tissue-based biomarker exploration. © 2019 International Society for Advancement of Cytometry.

FOXA1 overexpression mediates endocrine resistance by altering the ER transcriptome and IL-8 expression in ER-positive breast cancer.

Forkhead box protein A1 (FOXA1) is a pioneer factor of estrogen receptor α (ER)-chromatin binding and function, yet its aberration in endocrine-resistant (Endo-R) breast cancer is unknown. Here, we report preclinical evidence for a role of FOXA1 in Endo-R breast cancer as well as evidence for its clinical significance. FOXA1 is gene-amplified and/or overexpressed in Endo-R derivatives of several breast cancer cell line models. Induced FOXA1 triggers oncogenic gene signatures and proteomic profiles highly associated with endocrine resistance. Integrated omics data reveal IL8 as one of the most perturbed genes regulated by FOXA1 and ER transcriptional reprogramming in Endo-R cells. IL-8 knockdown inhibits tamoxifen-resistant cell growth and invasion and partially attenuates the effect of overexpressed FOXA1. Our study highlights a role of FOXA1 via IL-8 signaling as a potential therapeutic target in FOXA1-overexpressing ER-positive tumors.

Simultaneous Multicolor Single-Molecule Tracking with Single-Laser Excitation via Spectral Imaging.

Single-molecule tracking (SMT) offers rich information on the dynamics of underlying biological processes, but multicolor SMT has been challenging due to spectral cross talk and a need for multiple laser excitations. Here, we describe a single-molecule spectral imaging approach for live-cell tracking of multiple fluorescent species at once using a single-laser excitation. Fluorescence signals from all the molecules in the field of view are collected using a single objective and split between positional and spectral channels. Images of the same molecule in the two channels are then combined to determine both the location and the identity of the molecule. The single-objective configuration of our approach allows for flexible sample geometry and the use of a live-cell incubation chamber required for live-cell SMT. Despite a lower photon yield, we achieve excellent spatial (20-40 nm) and spectral (10-15 nm) resolutions comparable to those obtained with dual-objective, spectrally resolved Stochastic Optical Reconstruction Microscopy. Furthermore, motions of the fluorescent molecules did not cause loss of spectral resolution owing to the dual-channel spectral calibration. We demonstrate SMT in three (and potentially more) colors using spectrally proximal fluorophores and single-laser excitation, and show that trajectories of each species can be reliably extracted with minimal cross talk.

Implementing a comprehensive translational oncology platform: from molecular testing to actionability.

BACKGROUND: In order to establish the workflows required to implement a real-time process involving multi-omic analysis of patient samples to support precision-guided therapeutic intervention, a tissue acquisition and analysis trial was implemented. This report describes our findings to date, including the frequency with which mutational testing led to precision-guided therapy and outcome for those patients. METHODS: Eligible patients presenting to Oregon Health and Science University Knight Cancer Institute were enrolled on the study. Patients with biopsy proven metastatic or locally advanced unresectable prostate cancer, breast cancer, pancreatic adenocarcinoma, or refractory acute myelogenous leukemia receiving standard of care therapy were eligible. Metastatic site biopsies were collected and analyzed using the Knight Diagnostic Lab GeneTrails comprehensive solid tumor panel (124 genes). CLIA certified genomic information was made available to the treating physician. RESULTS: Between 1/26/2017 and 5/30/2018, 38 patients were enrolled, with 28 successfully undergoing biopsy. Of these, 25 samples yielded sufficient tumor for analysis. The median biopsy cellularity and number of cores collected were 70% (15-90%) and 5 (2-20), respectively. No procedure-related complications occurred. GeneTrails analysis revealed that 22 of 25 (88%) tumor samples harbored at least one potentially actionable mutation, and 18 (72%) samples harbored 2 or more potentially actionable mutations. The most common genetic alterations identified involved: DNA damage repair genes, cell cycle regulating genes, PIK3CA/Akt/mTOR pathway, and FGF gene family. To date, CLIA certified genomic results were used by treating physicians for precision-guided therapy in 5 (23%) patients. CONCLUSION: We report the feasibility of real-time tissue acquisition and analysis to support a successful translational oncology platform. The workflow will provide the foundation to improve access and accrual to biomarker driven precision oncology trials.

Ras-GTP dimers activate the Mitogen-Activated Protein Kinase (MAPK) pathway.

Rat sarcoma (Ras) GTPases regulate cell proliferation and survival through effector pathways including Raf-MAPK, and are the most frequently mutated genes in human cancer. Although it is well established that Ras activity requires binding to both GTP and the membrane, details of how Ras operates on the cell membrane to activate its effectors remain elusive. Efforts to target mutant Ras in human cancers to therapeutic benefit have also been largely unsuccessful. Here we show that Ras-GTP forms dimers to activate MAPK. We used quantitative photoactivated localization microscopy (PALM) to analyze the nanoscale spatial organization of PAmCherry1-tagged KRas 4B (hereafter referred to KRas) on the cell membrane under various signaling conditions. We found that at endogenous expression levels KRas forms dimers, and KRas(G12D), a mutant that constitutively binds GTP, activates MAPK. Overexpression of KRas leads to formation of higher order Ras nanoclusters. Conversely, at lower expression levels, KRas(G12D) is monomeric and activates MAPK only when artificially dimerized. Moreover, dimerization and signaling of KRas are both dependent on an intact CAAX (C, cysteine; A, aliphatic; X, any amino acid) motif that is also known to mediate membrane localization. These results reveal a new, dimerization-dependent signaling mechanism of Ras, and suggest Ras dimers as a potential therapeutic target in mutant Ras-driven tumors.

Phylogenetic analyses of melanoma reveal complex patterns of metastatic dissemination.

Melanoma is difficult to treat once it becomes metastatic. However, the precise ancestral relationship between primary tumors and their metastases is not well understood. We performed whole-exome sequencing of primary melanomas and multiple matched metastases from eight patients to elucidate their phylogenetic relationships. In six of eight patients, we found that genetically distinct cell populations in the primary tumor metastasized in parallel to different anatomic sites, rather than sequentially from one site to the next. In five of these six patients, the metastasizing cells had themselves arisen from a common parental subpopulation in the primary, indicating that the ability to establish metastases is a late-evolving trait. Interestingly, we discovered that individual metastases were sometimes founded by multiple cell populations of the primary that were genetically distinct. Such establishment of metastases by multiple tumor subpopulations could help explain why identical resistance variants are identified in different sites after initial response to systemic therapy. One primary tumor harbored two subclones with different oncogenic mutations in CTNNB1, which were both propagated to the same metastasis, raising the possibility that activation of wingless-type mouse mammary tumor virus integration site (WNT) signaling may be involved, as has been suggested by experimental models.

Characterizing the immune microenvironment in high-risk ductal carcinoma in situ of the breast.

PURPOSE: The recent increase in the incidence of ductal carcinoma in situ (DCIS) has sparked debate over the classification and treatment of this disease. Although DCIS is considered a precursor lesion to invasive breast cancer, some DCIS may have more or less risk than is realized. In this study, we characterized the immune microenvironment in DCIS to determine if immune infiltrates are predictive of recurrence. METHODS: Fifty-two cases of high-grade DCIS (HG-DCIS), enriched for large lesions and a history of recurrence, were age matched with 65 cases of non-high-grade DCIS (nHG-DCIS). Immune infiltrates were characterized by single- or dual-color staining of FFPE sections for the following antigens: CD4, CD8, CD20, FoxP3, CD68, CD115, Mac387, MRC1, HLA-DR, and PCNA. Nuance multispectral imaging software was used for image acquisition. Protocols for automated image analysis were developed using CellProfiler. Immune cell populations associated with risk of recurrence were identified using classification and regression tree analysis. RESULTS: HG-DCIS had significantly higher percentages of FoxP3+ cells, CD68+ and CD68+PCNA+ macrophages, HLA-DR+ cells, CD4+ T cells, CD20+ B cells, and total tumor infiltrating lymphocytes compared to nHG-DCIS. A classification tree, generated from 16 immune cell populations and 8 clinical parameters, identified three immune cell populations associated with risk of recurrence: CD8+HLADR+ T cells, CD8+HLADR- T cells, and CD115+ cells. CONCLUSION: These findings suggest that the tumor immune microenvironment is an important factor in identifying DCIS cases with the highest risk for recurrence and that manipulating the immune microenvironment may be an efficacious strategy to alter or prevent disease progression.

Genome co-amplification upregulates a mitotic gene network activity that predicts outcome and response to mitotic protein inhibitors in breast cancer.

BACKGROUND: High mitotic activity is associated with the genesis and progression of many cancers. Small molecule inhibitors of mitotic apparatus proteins are now being developed and evaluated clinically as anticancer agents. With clinical trials of several of these experimental compounds underway, it is important to understand the molecular mechanisms that determine high mitotic activity, identify tumor subtypes that carry molecular aberrations that confer high mitotic activity, and to develop molecular markers that distinguish which tumors will be most responsive to mitotic apparatus inhibitors. METHODS: We identified a coordinately regulated mitotic apparatus network by analyzing gene expression profiles for 53 malignant and non-malignant human breast cancer cell lines and two separate primary breast tumor datasets. We defined the mitotic network activity index (MNAI) as the sum of the transcriptional levels of the 54 coordinately regulated mitotic apparatus genes. The effect of those genes on cell growth was evaluated by small interfering RNA (siRNA). RESULTS: High MNAI was enriched in basal-like breast tumors and was associated with reduced survival duration and preferential sensitivity to inhibitors of the mitotic apparatus proteins, polo-like kinase, centromere associated protein E and aurora kinase designated GSK462364, GSK923295 and GSK1070916, respectively. Co-amplification of regions of chromosomes 8q24, 10p15-p12, 12p13, and 17q24-q25 was associated with the transcriptional upregulation of this network of 54 mitotic apparatus genes, and we identify transcription factors that localize to these regions and putatively regulate mitotic activity. Knockdown of the mitotic network by siRNA identified 22 genes that might be considered as additional therapeutic targets for this clinically relevant patient subgroup. CONCLUSIONS: We define a molecular signature which may guide therapeutic approaches for tumors with high mitotic network activity.

Endocytic trafficking of laminin is controlled by dystroglycan and is disrupted in cancers.

The dynamic interactions between cells and basement membranes serve as essential regulators of tissue architecture and function in metazoans, and perturbation of these interactions contributes to the progression of a wide range of human diseases, including cancers. Here, we reveal the pathway and mechanism for the endocytic trafficking of a prominent basement membrane protein, laminin-111 (referred to here as laminin), and their disruption in disease. Live-cell imaging of epithelial cells revealed pronounced internalization of laminin into endocytic vesicles. Laminin internalization was receptor mediated and dynamin dependent, and laminin proceeded to the lysosome through the late endosome. Manipulation of laminin receptor expression revealed that the dominant regulator of laminin internalization is dystroglycan, a laminin receptor that is functionally perturbed in muscular dystrophies and in many cancers. Correspondingly, laminin internalization was found to be deficient in aggressive cancer cells displaying non-functional dystroglycan, and restoration of dystroglycan function strongly enhanced the endocytosis of laminin in both breast cancer and glioblastoma cells. These results establish previously unrecognized mechanisms for the modulation of cell-basement-membrane communication in normal cells and identify a profound disruption of endocytic laminin trafficking in aggressive cancer subtypes.

Multiplexed imaging reveals heterogeneity of PI3K/MAPK network signaling in breast lesions of known PIK3CA genotype.

PURPOSE: Activating genetic changes in the phosphatidylinositol-3-kinase (PI3K) signaling pathway are found in over half of invasive breast cancers (IBCs). Previously, we discovered numerous hotspot PIK3CA mutations in proliferative breast lesions. Here, we investigate the spatial nature of PI3K pathway signaling and its relationship with PI3K genotype in breast lesions. METHODS: We identified PI3K phosphosignaling network signatures in columnar cell change (CCL), usual ductal hyperplasia (UDH), ductal carcinoma in situ (DCIS), and IBC in 26 lesions of known PIK3CA genotype from 10 human breast specimens using a hyperspectral-based multiplexed tissue imaging platform (MTIP) to simultaneously quantitate PI3K/MAPK pathway targets (pAKT473, pAKT308, pPRAS40, pS6, and pERK) in FFPE tissue, with single-cell resolution. RESULTS: We found that breast lesional epithelia contained spatially heterogeneous patterns of PI3K pathway phosphoprotein signatures, even within microscopic areas of CCL, UDH, DCIS, and IBC. Most lesions contained 3-12 unique phosphoprotein signatures within the same microscopic field. The dominant phosphoprotein signature for each lesion was not well correlated with lesion genotype or lesion histology, yet samples from the same patient tended to group together. Further, 5 UDH/CCL lesions across different patients had a common phosphosignature at the epithelial-stromal interface (possible myoepithelial cells) that was distinct from both the adjacent lesional epithelium and distinct from adjacent stroma. CONCLUSION: We present the first spatial mapping of PI3K phosphoprotein networks in proliferative breast lesions and demonstrate complex PI3K signaling heterogeneity that defies simple correlation between PIK3CA genotype and phosphosignal pattern.

Single-molecule superresolution imaging allows quantitative analysis of RAF multimer formation and signaling.

The RAF serine/threonine kinases regulate cell growth through the MAPK pathway, and are targeted by small-molecule RAF inhibitors (RAFis) in human cancer. It is now apparent that protein multimers play an important role in RAF activation and tumor response to RAFis. However, the exact stoichiometry and cellular location of these multimers remain unclear because of the lack of technologies to visualize them. In the present work, we demonstrate that photoactivated localization microscopy (PALM), in combination with quantitative spatial analysis, provides sufficient resolution to directly visualize protein multimers in cells. Quantitative PALM imaging showed that CRAF exists predominantly as cytoplasmic monomers under resting conditions but forms dimers as well as trimers and tetramers at the cell membrane in the presence of active RAS. In contrast, N-terminal truncated CRAF (CatC) lacking autoinhibitory domains forms constitutive dimers and occasional tetramers in the cytoplasm, whereas a CatC mutant with a disrupted CRAF-CRAF dimer interface does not. Finally, artificially forcing CRAF to the membrane by fusion to a RAS CAAX motif induces multimer formation but activates RAF/MAPK only if the dimer interface is intact. Together, these quantitative results directly confirm the existence of RAF dimers and potentially higher-order multimers and their involvement in cell signaling, and showed that RAF multimer formation can result from multiple mechanisms and is a critical but not sufficient step for RAF activation.

Cationic Polymer Modified Mesoporous Silica Nanoparticles for Targeted SiRNA Delivery to HER2+ Breast Cancer.

In vivo delivery of siRNAs designed to inhibit genes important in cancer and other diseases continues to be an important biomedical goal. We now describe a new nanoparticle construct that has been engineered for efficient delivery of siRNA to tumors. The construct is comprised of a 47-nm mesoporous silica nanoparticle (MSNP) core coated with a cross-linked PEI-PEG copolymer, carrying siRNA against the HER2 oncogene, and coupled to the anti-HER2 monoclonal antibody (trastuzumab). The construct has been engineered to increase siRNA blood half-life, enhance tumor-specific cellular uptake, and maximize siRNA knockdown efficacy. The optimized anti-HER2-nanoparticles produced apoptotic death in HER2 positive (HER2+) breast cancer cells grown in vitro, but not in HER2 negative (HER2-) cells. One dose of the siHER2-nanoparticles reduced HER2 protein levels by 60% in trastuzumab-resistant HCC1954 xenografts. Multiple doses administered intravenously over 3 weeks significantly inhibited tumor growth (p < 0.004). The siHER2-nanoparticles have an excellent safety profile in terms of blood compatibility and low cytokine induction, when exposed to human peripheral blood mononuclear cells. The construct can be produced with high batch-to-batch reproducibility and the production methods are suitable for large-scale production. These results suggest that this siHER2-nanoparticle is ready for clinical evaluation.

Causal network inference using biochemical kinetics.

MOTIVATION: Networks are widely used as structural summaries of biochemical systems. Statistical estimation of networks is usually based on linear or discrete models. However, the dynamics of biochemical systems are generally non-linear, suggesting that suitable non-linear formulations may offer gains with respect to causal network inference and aid in associated prediction problems. RESULTS: We present a general framework for network inference and dynamical prediction using time course data that is rooted in non-linear biochemical kinetics. This is achieved by considering a dynamical system based on a chemical reaction graph with associated kinetic parameters. Both the graph and kinetic parameters are treated as unknown; inference is carried out within a Bayesian framework. This allows prediction of dynamical behavior even when the underlying reaction graph itself is unknown or uncertain. Results, based on (i) data simulated from a mechanistic model of mitogen-activated protein kinase signaling and (ii) phosphoproteomic data from cancer cell lines, demonstrate that non-linear formulations can yield gains in causal network inference and permit dynamical prediction and uncertainty quantification in the challenging setting where the reaction graph is unknown. AVAILABILITY AND IMPLEMENTATION: MATLAB R2014a software is available to download from warwick.ac.uk/chrisoates. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.