Extracting Insights From Temporal Data by Integrating Dynamic Modeling and Machine Learning.

Biological processes are dynamic. As a result, temporal analyses are necessary to fully understand the complex interactions that occurs within these systems. One example of a multifaceted biological process is restitution: the initial step in complex wound repair. Restitution is a dynamic process that depends on an elegant orchestration between damaged cells and their intact neighbors. Such orchestration enables the quick repair of the damaged area, which is essential to preserve epithelial integrity and prevent further injury. High quality dynamic data of the cellular and molecular events that make up the gastric restitution process has been documented. However, comprehensive dynamic models that connect all relevant molecular interactions to cellular behaviors are challenging to construct and experimentally validate. In order to efficiently provide feedback to ongoing experimental work, we have integrated dynamical modeling and machine learning to efficiently extract data-driven insights without incorporating detailed mechanisms. Dynamical models convert time course data into a set of static features, which are then subjected to machine learning analysis. The integrated analysis provides data-driven insights into how repair might be regulated in individual gastric organoids. We have provided a "proof of concept" of how such an analysis pipeline can be used to analyze any temporal dataset and provide timely data-driven insights.

Designing combination therapies with modeling chaperoned machine learning.

Chemotherapy resistance is a major challenge to the effective treatment of cancer. Thus, a systematic pipeline for the efficient identification of effective combination treatments could bring huge biomedical benefit. In order to facilitate rational design of combination therapies, we developed a comprehensive computational model that incorporates the available biological knowledge and relevant experimental data on the life-and-death response of individual cancer cells to cisplatin or cisplatin combined with the TNF-related apoptosis-inducing ligand (TRAIL). The model's predictions, that a combination treatment of cisplatin and TRAIL would enhance cancer cell death and exhibit a "two-wave killing" temporal pattern, was validated by measuring the dynamics of p53 accumulation, cell fate, and cell death in single cells. The validated model was then subjected to a systematic analysis with an ensemble of diverse machine learning methods. Though each method is characterized by a different algorithm, they collectively identified several molecular players that can sensitize tumor cells to cisplatin-induced apoptosis (sensitizers). The identified sensitizers are consistent with previous experimental observations. Overall, we have illustrated that machine learning analysis of an experimentally validated mechanistic model can convert our available knowledge into the identity of biologically meaningful sensitizers. This knowledge can then be leveraged to design treatment strategies that could improve the efficacy of chemotherapy.

Correction to: A quantitative systems pharmacology (QSP) model for Pneumocystis treatment in mice.

It was highlighted that the original article [1] contained errors in the figures and their legends and by extension the in-text figure citations. This Corrections article shows the correct figures and correct figure legends.

Unraveling the Control of Cell Cycle Periods during Intestinal Stem Cell Differentiation.

During differentiation, intestinal stem cells (ISCs), a prototypical adult stem cell pool, become either secretory transit-amplifying cells, which give rise to all secretory cell types, or absorptive transit-amplifying cells, which give rise to enterocytes. These cells exhibit distinct cell cycle dynamics: ISCs cycle with a period of 24 h and absorptive transit-amplifying cells cycle with a period of ∼12 h, whereas secretory transit-amplifying cells arrest their cycle. The cell cycle dynamics of ISCs and their progeny are a systems-level property that emerges from interactions between the cell cycle control machinery and multiple regulatory pathways. Although many mathematical models have been developed to study the details of the cell cycle and related regulatory pathways, few models have been constructed to unravel the dynamic consequences of their interactions. To fill this gap, we present a simplified model focusing on the interaction between four key regulatory pathways (STAT, Wnt, Notch, and MAPK) and cell cycle control. After experimentally validating a model prediction, which showed that the Notch pathway can fine-tune the cell cycle period, we perform further model analysis that reveals that the change of cell cycle period accompanying ISC differentiation may be controlled by a design principle that has been well studied in dynamical systems theory-a saddle node on invariant circle bifurcation. Given that the mechanisms that control the cell cycle are conserved in most eukaryotic cell types, this general principle potentially controls the interplay between proliferation and differentiation for a broad range of stem cells.

A quantitative systems pharmacology (QSP) model for Pneumocystis treatment in mice.

BACKGROUND: The yeast-like fungi Pneumocystis, resides in lung alveoli and can cause a lethal infection known as Pneumocystis pneumonia (PCP) in hosts with impaired immune systems. Current therapies for PCP, such as trimethoprim-sulfamethoxazole (TMP-SMX), suffer from significant treatment failures and a multitude of serious side effects. Novel therapeutic approaches (i.e. newly developed drugs or novel combinations of available drugs) are needed to treat this potentially lethal opportunistic infection. Quantitative Systems Pharmacological (QSP) models promise to aid in the development of novel therapies by integrating available pharmacokinetic (PK) and pharmacodynamic (PD) knowledge to predict the effects of new treatment regimens. RESULTS: In this work, we constructed and independently validated PK modules of a number of drugs with available pharmacokinetic data. Characterized by simple structures and well constrained parameters, these PK modules could serve as a convenient tool to summarize and predict pharmacokinetic profiles. With the currently accepted hypotheses on the life stages of Pneumocystis, we also constructed a PD module to describe the proliferation, transformation, and death of Pneumocystis. By integrating the PK module and the PD module, the QSP model was constrained with observed levels of asci and trophic forms following treatments with multiple drugs. Furthermore, the temporal dynamics of the QSP model were validated with corresponding data. CONCLUSIONS: We developed and validated a QSP model that integrates available data and promises to facilitate the design of future therapies against PCP.

Multiscale positive feedbacks contribute to unidirectional gastric disease progression induced by helicobacter pylori infection.

BACKGROUND: Helicobacter Pylori (HP) is the most common risk factor for gastric cancer. Nearly half the world's population is infected with HP, but only a small percentage of those develop significant pathology. The bacteria itself does not directly cause cancer; rather it promotes an environment that is conducive to tumor formation. Upon infection, HP induces transcriptional changes in the host, leading to enhanced proliferation and host immune response. In addition, HP causes direct damage to gastric epithelial cells. RESULTS: We present a multiscale mechanistic model of HP induced changes. The model includes four modules representing the host transcriptional changes in response to infection, gastric atrophy, the Hedgehog pathway response, and the restriction point that controls cell cycle. This model was able to recapture a number of literature reported observations and was used as an "in silico" representation of the biological system for further analysis. Dynamical analysis of the model revealed that HP might induce the activation of multiple interplayed positive feedbacks, which in turn might result in a "ratchet ladder" system that promotes a unidirectional progression of gastric disease. CONCLUSIONS: The current multiscale model is able to recapitulate the observed experimental features of HP host interactions and provides dynamic insights on the epidemiologically observed heterogeneity in disease progression. This model provides a solid framework that can be further expanded and validated to include additional experimental evidence, to understand the complex multi-pathway interactions characterizing HP infection, and to design novel treatment protocols for HP induced diseases.

A dynamical framework for complex fractional killing.

When chemotherapy drugs are applied to tumor cells with the same or similar genotypes, some cells are killed, while others survive. This fractional killing contributes to drug resistance in cancer. Through an incoherent feedforward loop, chemotherapy drugs not only activate p53 to induce cell death, but also promote the expression of apoptosis inhibitors which inhibit cell death. Consequently, cells in which p53 is activated early undergo apoptosis while cells in which p53 is activated late survive. The incoherent feedforward loop and the essential role of p53 activation timing makes fractional killing a complex dynamical challenge, which is hard to understand with intuition alone. To better understand this process, we have constructed a representative model by integrating the control of apoptosis with the relevant signaling pathways. After the model was trained to recapture the observed properties of fractional killing, it was analyzed with nonlinear dynamical tools. The analysis suggested a simple dynamical framework for fractional killing, which predicts that cell fate can be altered in three possible ways: alteration of bifurcation geometry, alteration of cell trajectories, or both. These predicted categories can explain existing strategies known to combat fractional killing and facilitate the design of novel strategies.

A fast and robust protocol for metataxonomic analysis using RNAseq data.

BACKGROUND: Metagenomics is a rapidly emerging field aimed to analyze microbial diversity and dynamics by studying the genomic content of the microbiota. Metataxonomics tools analyze high-throughput sequencing data, primarily from 16S rRNA gene sequencing and DNAseq, to identify microorganisms and viruses within a complex mixture. With the growing demand for analysis of the functional microbiome, metatranscriptome studies attract more interest. To make metatranscriptomic data sufficient for metataxonomics, new analytical workflows are needed to deal with sparse and taxonomically less informative sequencing data. RESULTS: We present a new protocol, IMSA+A, for accurate taxonomy classification based on metatranscriptome data of any read length that can efficiently and robustly identify bacteria, fungi, and viruses in the same sample. The new protocol improves accuracy by using a conservative reference database, employing a new counting scheme, and by assembling shotgun reads. Assembly also reduces analysis runtime. Simulated data were utilized to evaluate the protocol by permuting common experimental variables. When applied to the real metatranscriptome data for mouse intestines colonized by ASF, the protocol showed superior performance in detection of the microorganisms compared to the existing metataxonomics tools. IMSA+A is available at https://github.com/JeremyCoxBMI/IMSA-A . CONCLUSIONS: The developed protocol addresses the need for taxonomy classification from RNAseq data. Previously not utilized, i.e., unmapped to a reference genome, RNAseq reads can now be used to gather taxonomic information about the microbiota present in a biological sample without conducting additional sequencing. Any metatranscriptome pipeline that includes assembly of reads can add this analysis with minimal additional cost of compute time. The new protocol also creates an opportunity to revisit old metatranscriptome data, where taxonomic content may be important but was not analyzed.