A Bayesian approach to determine the composition of heterogeneous cancer tissue.

BACKGROUND: Cancer Tissue Heterogeneity is an important consideration in cancer research as it can give insights into the causes and progression of cancer. It is known to play a significant role in cancer cell survival, growth and metastasis. Determining the compositional breakup of a heterogeneous cancer tissue can also help address the therapeutic challenges posed by heterogeneity. This necessitates a low cost, scalable algorithm to address the challenge of accurate estimation of the composition of a heterogeneous cancer tissue. METHODS: In this paper, we propose an algorithm to tackle this problem by utilizing the data of accurate, but high cost, single cell line cell-by-cell observation methods in low cost aggregate observation method for heterogeneous cancer cell mixtures to obtain their composition in a Bayesian framework. RESULTS: The algorithm is analyzed and validated using synthetic data and experimental data. The experimental data is obtained from mixtures of three separate human cancer cell lines, HCT116 (Colorectal carcinoma), A2058 (Melanoma) and SW480 (Colorectal carcinoma). CONCLUSION: The algorithm provides a low cost framework to determine the composition of heterogeneous cancer tissue which is a crucial aspect in cancer research.

Identifying mechanistic similarities in drug responses.

MOTIVATION: In early drug development, it would be beneficial to be able to identify those dynamic patterns of gene response that indicate that drugs targeting a particular gene will be likely or not to elicit the desired response. One approach would be to quantitate the degree of similarity between the responses that cells show when exposed to drugs, so that consistencies in the regulation of cellular response processes that produce success or failure can be more readily identified. RESULTS: We track drug response using fluorescent proteins as transcription activity reporters. Our basic assumption is that drugs inducing very similar alteration in transcriptional regulation will produce similar temporal trajectories on many of the reporter proteins and hence be identified as having similarities in their mechanisms of action (MOA). The main body of this work is devoted to characterizing similarity in temporal trajectories/signals. To do so, we must first identify the key points that determine mechanistic similarity between two drug responses. Directly comparing points on the two signals is unrealistic, as it cannot handle delays and speed variations on the time axis. Hence, to capture the similarities between reporter responses, we develop an alignment algorithm that is robust to noise, time delays and is able to find all the contiguous parts of signals centered about a core alignment (reflecting a core mechanism in drug response). Applying the proposed algorithm to a range of real drug experiments shows that the result agrees well with the prior drug MOA knowledge. AVAILABILITY: The R code for the RLCSS algorithm is available at http://gsp.tamu.edu/Publications/supplementary/zhao12a.

Assessing the efficacy of molecularly targeted agents on cell line-based platforms by using system identification.

BACKGROUND: Molecularly targeted agents (MTAs) are increasingly used for cancer treatment, the goal being to improve the efficacy and selectivity of cancer treatment by developing agents that block the growth of cancer cells by interfering with specific targeted molecules needed for carcinogenesis and tumor growth. This approach differs from traditional cytotoxic anticancer drugs. The lack of specificity of cytotoxic drugs allows a relatively straightforward approach in preclinical and clinical studies, where the optimal dose has usually been defined as the "maximum tolerated dose" (MTD). This toxicity-based dosing approach is founded on the assumption that the therapeutic anticancer effect and toxic effects of the drug increase in parallel as the dose is escalated. On the contrary, most MTAs are expected to be more selective and less toxic than cytotoxic drugs. Consequently, the maximum therapeutic effect may be achieved at a "biologically effective dose" (BED) well below the MTD. Hence, dosing study for MTAs should be different from cytotoxic drugs. Enhanced efforts to molecularly characterize the drug efficacy for MTAs in preclinical models will be valuable for successfully designing dosing regimens for clinical trials. RESULTS: A novel preclinical model combining experimental methods and theoretical analysis is proposed to investigate the mechanism of action and identify pharmacodynamic characteristics of the drug. Instead of fixed time point analysis of the drug exposure to drug effect, the time course of drug effect for different doses is quantitatively studied on cell line-based platforms using system identification, where tumor cells' responses to drugs through the use of fluorescent reporters are sampled over a time course. Results show that drug effect is time-varying and higher dosages induce faster and stronger responses as expected. However, the drug efficacy change along different dosages is not linear; on the contrary, there exist certain thresholds. This kind of preclinical study can provide valuable suggestions about dosing regimens for the in vivo experimental stage to increase productivity.

In Silico Modeling of the Induction of Apoptosis by Cryptotanshinone in Osteosarcoma Cell Lines.

Osteosarcoma (OS) is the most common primary malignant bone tumor of both children and pet canines. Its characteristic genomic instability and complexity coupled with the dearth of knowledge about its etiology has made improvement in the current treatment difficult. We use the existing literature about the biological pathways active in OS and combine it with the current research involving natural compounds to identify new targets and design more effective drug therapies. The key components of these pathways are modeled as a Boolean network with multiple inputs and multiple outputs. The combinatorial circuit is employed to theoretically predict the efficacies of various drugs in combination with Cryptotanshinone. We show that the action of the herbal drug, Cryptotanshinone on OS cell lines induces apoptosis by increasing sensitivity to TNF-related apoptosis-inducing ligand (TRAIL) through its multi-pronged action on STAT3, DRP1 and DR5. The Boolean framework is used to detect additional drug intervention points in the pathway that could amplify the action of Cryptotanshinone.

Comprehensive live-cell imaging analysis of cryptotanshinone and synergistic drug-screening effects in various human and canine cancer cell lines.

BACKGROUND: Several studies have highlighted both the extreme anticancer effects of Cryptotanshinone (CT), a Stat3 crippling component from Salvia miltiorrhiza, as well as other STAT3 inhibitors to fight cancer. METHODS: Data presented in this experiment incorporates 2 years of in vitro studies applying a comprehensive live-cell drug-screening analysis of human and canine cancer cells exposed to CT at 20 µM concentration, as well as to other drug combinations. As previously observed in other studies, dogs are natural cancer models, given to their similarity in cancer genetics, epidemiology and disease progression compared to humans. RESULTS: Results obtained from several types of human and canine cancer cells exposed to CT and varied drug combinations, verified CT efficacy at combating cancer by achieving an extremely high percentage of apoptosis within 24 hours of drug exposure. CONCLUSIONS: CT anticancer efficacy in various human and canine cancer cell lines denotes its ability to interact across different biological processes and cancer regulatory cell networks, driving inhibition of cancer cell survival.

A Gaussian Mixture-Model Exploiting Pathway Knowledge for Dissecting Cancer Heterogeneity.

In this work, we develop a systematic approach for applying pathway knowledge to a multivariate Gaussian mixture model for dissecting a heterogeneous cancer tissue. The downstream transcription factors are selected as observables from available partial pathway knowledge in such a way that the subpopulations produce some differential behavior in response to the drugs selected in the upstream. For each subpopulation, each unique (drug, observable) pair is considered as a unique dimension of a multivariate Gaussian distribution. Expectation-maximization (EM) algorithm with hill-climbing is then used to rank the most probable estimates of the mixture composition based on the log-likelihood value. A major contribution of this work is to examine the efficacy of the EM based approach in estimating the composition of experimental mixture sets from cell-by-cell measurements collected on a dynamic cell imaging platform. Towards this end, we apply the algorithm on hourly data collected for two different mixture compositions of A2058, HCT116, and SW480 cell lines for three scenarios: untreated, Lapatinib-treated, and Temsirolimus-treated. Additionally, we show how this methodology can provide a basis for comparing the killing rate of different drugs for a heterogeneous cancer tissue. This obviously has important implications for designing efficient drugs for treating heterogeneous malignant tumors.

Phenotype Classification Using Moment Features of Single-Cell Data.

Features for standard expression microarray and RNA-Seq classification are expression averages over collections of cells. Single cell provides expression measurements for individual cells in a collection of cells from a particular tissue sample. Hence, it can yield feature vectors consisting of higher order and mixed moments. This article demonstrates the advantage of using these expression moments in cancer-related classification. We use synthetic data generated from 2 real networks, the mammalian cell cycle network and a melanoma-related pathway network, and real single-cell data generated via fluorescent protein reporters from 2 cell lines, HT-29 and HCT-116. The networks consist of hidden binary regulatory networks with Gaussian observations. The steady-state distributions of both the original and mutated networks are found, and data are drawn from these for moment-based classification using the mean, variance, skewness, and mixed moments. For the real data, we only observe 1 gene at a time, so that only the mean, variance, and skewness are considered, the analysis being done for 2 genes, EGFR and ERRB2. For the synthetic data, classification improves as we move from just the mean to mean, variance, and skewness and then to these plus the mixed moments. Comparisons are done with 3, 4, or 5 features, using feature selection. Sample size effects are considered. For the real data, we only consider mean, variance, and skewness, with results improving when the higher order moments are used as features.

Validation of computational methods in genomics.

High-throughput technologies for genomics provide tens of thousands of genetic measurements, for instance, gene-expression measurements on microarrays, and the availability of these measurements has motivated the use of machine learning (inference) methods for classification, clustering, and gene networks. Generally, a design method will yield a model that satisfies some model constraints and fits the data in some manner. On the other hand, a scientific theory consists of two parts: (1) a mathematical model to characterize relations between variables, and (2) a set of relations between model variables and observables that are used to validate the model via predictive experiments. Although machine learning algorithms are constructed to hopefully produce valid scientific models, they do not ipso facto do so. In some cases, such as classifier estimation, there is a well-developed error theory that relates to model validity according to various statistical theorems, but in others such as clustering, there is a lack of understanding of the relationship between the learning algorithms and validation. The issue of validation is especially problematic in situations where the sample size is small in comparison with the dimensionality (number of variables), which is commonplace in genomics, because the convergence theory of learning algorithms is typically asymptotic and the algorithms often perform in counter-intuitive ways when used with samples that are small in relation to the number of variables. For translational genomics, validation is perhaps the most critical issue, because it is imperative that we understand the performance of a diagnostic or therapeutic procedure to be used in the clinic, and this performance relates directly to the validity of the model behind the procedure. This paper treats the validation issue as it appears in two classes of inference algorithms relating to genomics - classification and clustering. It formulates the problem and reviews salient results.

Systems biology for organotypic cell cultures.

Translating in vitro biological data into actionable information related to human health holds the potential to improve disease treatment and risk assessment of chemical exposures. While genomics has identified regulatory pathways at the cellular level, translation to the organism level requires a multiscale approach accounting for intra-cellular regulation, inter-cellular interaction, and tissue/organ-level effects. Tissue-level effects can now be probed in vitro thanks to recently developed systems of three-dimensional (3D), multicellular, "organotypic" cell cultures, which mimic functional responses of living tissue. However, there remains a knowledge gap regarding interactions across different biological scales, complicating accurate prediction of health outcomes from molecular/genomic data and tissue responses. Systems biology aims at mathematical modeling of complex, non-linear biological systems. We propose to apply a systems biology approach to achieve a computational representation of tissue-level physiological responses by integrating empirical data derived from organotypic culture systems with computational models of intracellular pathways to better predict human responses. Successful implementation of this integrated approach will provide a powerful tool for faster, more accurate and cost-effective screening of potential toxicants and therapeutics. On September 11, 2015, an interdisciplinary group of scientists, engineers, and clinicians gathered for a workshop in Research Triangle Park, North Carolina, to discuss this ambitious goal. Participants represented laboratory-based and computational modeling approaches to pharmacology and toxicology, as well as the pharmaceutical industry, government, non-profits, and academia. Discussions focused on identifying critical system perturbations to model, the computational tools required, and the experimental approaches best suited to generating key data.

Stress-specific signatures: expression profiling of p53 wild-type and -null human cells.

Gene expression responses of human cell lines exposed to a diverse set of stress agents were compared by cDNA microarray hybridization. The B-lymphoblastoid cell line TK6 (p53 wild-type) and its p53-null derivative, NH32, were treated in parallel to facilitate investigation of p53-dependent responses. RNA was extracted 4 h after the beginning of treatment when no notable decrease in cell viability was evident in the cultures. Gene expression signatures were defined that discriminated between four broad general mechanisms of stress agents: Non-DNA-damaging stresses (heat shock, osmotic shock, and 12-O-tetradecanoylphorbol 13-acetate), agents causing mainly oxidative stress (arsenite and hydrogen peroxide), ionizing radiations (neutron and gamma-ray exposures), and other DNA-damaging agents (ultraviolet radiation, methyl methanesulfonate, adriamycin, camptothecin, and cis-Platinum(II)diammine dichloride (cisplatin)). Within this data set, non-DNA-damaging stresses could be discriminated from all DNA-damaging stresses, and profiles for individual agents were also defined. While DNA-damaging stresses showed a strong p53-dependent element in their responses, no discernible p53-dependent responses were triggered by the non-DNA-damaging stresses. A set of 16 genes did exhibit a robust p53-dependent pattern of induction in response to all nine DNA-damaging agents, however.

Differential responses of stress genes to low dose-rate gamma irradiation.

In the past, most mechanistic studies of ionizing radiation response have employed very large doses, then extrapolated the results down to doses relevant to human exposure. It is becoming increasingly apparent, however, that this does not give an accurate or complete picture of the effects of most environmental exposures, which tend to be of low dose and protracted over time. We have initiated direct studies of low dose exposures, and using the relatively responsive ML-1 cell line, have shown that changes in gene expression can be triggered by doses of gamma-rays of 10 cGy and less in human cells. We have now extended these studies to investigate the effects on gene induction of reducing the rate of irradiation. In the ML-1 human myeloid leukemia cell line, we have found that reducing the dose rate over three orders of magnitude results in some protection against the induction of apoptosis, but still causes linear induction of the p53-regulated genes CDKN1A, GADD45A, and MDM2 between 2 and 50 cGy. Reducing the rate of exposure reduces the magnitude of induction of CDKN1A and GADD45A, but not the magnitude or duration of cell cycle delay. In contrast, MDM2 is induced to the same extent regardless of the rate of dose delivery. Microarray analysis has identified additional low dose-rate-inducible genes, and indicates the existence of two general classes of low dose-rate responders in ML-1. One group of genes is induced in a dose rate-dependent fashion, similar to GADD45A and CDKN1A. Functional annotation of this gene cluster indicates a preponderance of genes with known roles in apoptosis regulation. Similarly, a group of genes with dose rate-independent induction, such as seen for MDM2, was also identified. The majority of genes in this group are involved in cell cycle regulation. This apparent differential regulation of stress signaling pathways and outcomes in response to protracted radiation exposure has implications for carcinogenesis and risk assessment, and could not have been predicted from classical high dose studies.

Assessing Combinational Drug Efficacy in Cancer Cells by Using Image-based Dynamic Response Analysis.

The landscape of translational research has been shifting toward drug combination therapies. Pairing of drugs allows for more types of drug interaction with cells. In order to accurately and comprehensively assess combinational drug efficacy, analytical methods capable of recognizing these alternative reactions will be required to prioritize those drug candidates having better chances of delivering appreciable therapeutic benefits. Traditional efficacy measures are primarily based on the "extent" of drug inhibition, which is the percentage of cells being killed after drug exposure. Here, we introduce a second dimension of evaluation criterion, speed of killing, based on a live cell imaging assay. This dynamic response trajectory approach takes advantage of both "extent" and "speed" information and uncovers synergisms that would otherwise be missed, while also generating hypotheses regarding important mechanistic modes of drug action.

Strong feature sets from small samples.

For small samples, classifier design algorithms typically suffer from overfitting. Given a set of features, a classifier must be designed and its error estimated. For small samples, an error estimator may be unbiased but, owing to a large variance, often give very optimistic estimates. This paper proposes mitigating the small-sample problem by designing classifiers from a probability distribution resulting from spreading the mass of the sample points to make classification more difficult, while maintaining sample geometry. The algorithm is parameterized by the variance of the spreading distribution. By increasing the spread, the algorithm finds gene sets whose classification accuracy remains strong relative to greater spreading of the sample. The error gives a measure of the strength of the feature set as a function of the spread. The algorithm yields feature sets that can distinguish the two classes, not only for the sample data, but for distributions spread beyond the sample data. For linear classifiers, the topic of the present paper, the classifiers are derived analytically from the model, thereby providing an enormous savings in computation time. The algorithm is applied to cancer classification via cDNA microarrays. In particular, the genes BRCA1 and BRCA2 are associated with a hereditary disposition to breast cancer, and the algorithm is used to find gene sets whose expressions can be used to classify BRCA1 and BRCA2 tumors.

Simulation of cDNA microarrays via a parameterized random signal model.

cDNA microarrays provide simultaneous expression measurements for thousands of genes that are the result of processing images to recover the average signal intensity from a spot composed of pixels covering the area upon which the cDNA detector has been put down. The accuracy of the signal measurement depends on using an appropriate algorithm to process the images. This includes determining spot locations and processing the data in such a way as to take into account spot geometry, background noise, and various kinds of noise that degrade the signal. This paper presents a stochastic model for microarray images. There are over 20 model parameters, each governed by a probability distribution, that control the signal intensity, spot geometry, spot drift, background effects, and the many kinds of noise that affect microarray images owing to the manner in which they are formed. The model can be used to analyze the performance of image algorithms designed to measure the true signal intensity because the ground truth (signal intensity) for each spot is known. The levels of foreground noise, background noise, and spot distortion can be set, and algorithms can be evaluated under varying conditions.

Noise factor analysis for cDNA microarrays.

A microarray-image model is used that takes into account many factors, including spot morphology, signal strength, background fluorescent noise, and shape and surface degradation. The model yields synthetic images whose appearance and quality reflect that of real microarray images. The model is used to link noise factors to the fidelity of signal extraction with respect to a standard image-extraction algorithm. Of particular interest is the identification of the noise factors and their interactions that significantly degrade the ability to accurately detect the true gene-expression signal. This study uses statistical criteria in conjunction with the simulation of various noise conditions to better understand the noise influence on signal extraction for cDNA microarray images. It proposes a paradigm that is implemented in software. It specifically considers certain kinds of noise in the noise model and sets these at certain levels; however, one can choose other types of noise or use different noise levels. In sum, it develops a statistical package that can work in conjunction with the existing image simulation toolbox.

Functional genomics of UV radiation responses in human cells.

The gene expression responses of MCF-7, a p53 wild-type (wt) human cell line, were monitored by cDNA microarray hybridization after exposure to different wavelengths of UV irradiation. Equitoxic doses of UVA, UVB, and UVC radiation were used to reduce survival to 37%. The effects of suramin, a signal pathway inhibitor, on the gene expression responses to the three UV wavelengths were also compared in this model system. UVB radiation triggered the broadest gene expression responses, and 172 genes were found to be consistently responsive in at least two-thirds of independent UVB experiments. These UVB radiation-responsive genes encode proteins with diverse cellular roles including cell cycle control, DNA repair, signaling, transcription, protein synthesis, protein degradation, and RNA metabolism. The set of UVB-responsive genes included most of the genes responding to an equitoxic dose of UVC radiation, plus additional genes that were not strongly triggered by UVC radiation. There was also some overlap with genes responding to an equitoxic dose of UVA radiation, although responses to this lower energy UV radiation were overall weaker. Signaling through growth factor receptors and other cytokine receptors was shown to have a major role in mediating UV radiation stress responses, as suramin, which inhibits such receptors, attenuated responses to UV radiation in nearly all the cases. Inhibition by suramin was greater for UVC than for UVB irradiation. This probably reflects the more prominent role in UVB damage response of signaling by reactive oxygen species, which would not be affected by suramin. Our results with suramin demonstrate the power of cDNA microarray hybridization to illuminate the global effects of a pharmacologic inhibitor on cell signaling.

Evaluating gene set enrichment analysis via a hybrid data model.

Gene set enrichment analysis (GSA) methods have been widely adopted by biological labs to analyze data and generate hypotheses for validation. Most of the existing comparison studies focus on whether the existing GSA methods can produce accurate P-values; however, practitioners are often more concerned with the correct gene-set ranking generated by the methods. The ranking performance is closely related to two critical goals associated with GSA methods: the ability to reveal biological themes and ensuring reproducibility, especially for small-sample studies. We have conducted a comprehensive simulation study focusing on the ranking performance of seven representative GSA methods. We overcome the limitation on the availability of real data sets by creating hybrid data models from existing large data sets. To build the data model, we pick a master gene from the data set to form the ground truth and artificially generate the phenotype labels. Multiple hybrid data models can be constructed from one data set and multiple data sets of smaller sizes can be generated by resampling the original data set. This approach enables us to generate a large batch of data sets to check the ranking performance of GSA methods. Our simulation study reveals that for the proposed data model, the Q2 type GSA methods have in general better performance than other GSA methods and the global test has the most robust results. The properties of a data set play a critical role in the performance. For the data sets with highly connected genes, all GSA methods suffer significantly in performance.

Newton, laplace, and the epistemology of systems biology.

For science, theoretical or applied, to significantly advance, researchers must use the most appropriate mathematical methods. A century and a half elapsed between Newton's development of the calculus and Laplace's development of celestial mechanics. One cannot imagine the latter without the former. Today, more than three-quarters of a century has elapsed since the birth of stochastic systems theory. This article provides a perspective on the utilization of systems theory as the proper vehicle for the development of systems biology and its application to complex regulatory diseases such as cancer.

Tracking transcriptional activities with high-content epifluorescent imaging.

High-content cell imaging based on fluorescent protein reporters has recently been used to track the transcriptional activities of multiple genes under different external stimuli for extended periods. This technology enhances our ability to discover treatment-induced regulatory mechanisms, temporally order their onsets and recognize their relationships. To fully realize these possibilities and explore their potential in biological and pharmaceutical applications, we introduce a new data processing procedure to extract information about the dynamics of cell processes based on this technology. The proposed procedure contains two parts: (1) image processing, where the fluorescent images are processed to identify individual cells and allow their transcriptional activity levels to be quantified; and (2) data representation, where the extracted time course data are summarized and represented in a way that facilitates efficient evaluation. Experiments show that the proposed procedure achieves fast and robust image segmentation with sufficient accuracy. The extracted cellular dynamics are highly reproducible and sensitive enough to detect subtle activity differences and identify mechanisms responding to selected perturbations. This method should be able to help biologists identify the alterations of cellular mechanisms that allow drug candidates to change cell behavior and thereby improve the efficiency of drug discovery and treatment design.

Characterization of drug efficacy regions based on dosage and frequency schedules.

This paper proposes a framework to study the drug effect at the molecular level in order to address the following question of current interest in the drug community: Given a fixed total delivered drug, which is better, frequent small or infrequent large drug dosages? A hybrid system model is proposed to link the drug's pharmacokinetic and pharmacodynamic information, and allows the drug effects for different dosages and treatment schedules to be compared. A hybrid model facilitates the modeling of continuous quantitative changes that leads to discrete transitions. An optimal dosage-frequency regimen and the necessary and sufficient conditions for the drug to be effective are obtained analytically when the drug is designed to control a target gene. Then, we extend the analysis to the case where the target gene is part of a genetic regulatory network. A crucial observation is that there exists a "sweet spot," defined as the "drug efficacy region (DER)" in this paper, for certain dosage and frequency arrangements given the total delivered drug. This paper quantifies the therapeutic benefits of dosage regimen lying within the DER. Simulations are performed using MATLAB/SIMULINK to validate the analytical results.