Machine learning phenomics (MLP) combining deep learning with time-lapse-microscopy for monitoring colorectal adenocarcinoma cells gene expression and drug-response.

High-throughput phenotyping is becoming increasingly available thanks to analytical and bioinformatics approaches that enable the use of very high-dimensional data and to the availability of dynamic models that link phenomena across levels: from genes to cells, from cells to organs, and through the whole organism. The combination of phenomics, deep learning, and machine learning represents a strong potential for the phenotypical investigation, leading the way to a more embracing approach, called machine learning phenomics (MLP). In particular, in this work we present a novel MLP platform for phenomics investigation of cancer-cells response to therapy, exploiting and combining the potential of time-lapse microscopy for cell behavior data acquisition and robust deep learning software architectures for the latent phenotypes extraction. A two-step proof of concepts is designed. First, we demonstrate a strict correlation among gene expression and cell phenotype with the aim to identify new biomarkers and targets for tailored therapy in human colorectal cancer onset and progression. Experiments were conducted on human colorectal adenocarcinoma cells (DLD-1) and their profile was compared with an isogenic line in which the expression of LOX-1 transcript was knocked down. In addition, we also evaluate the phenotypic impact of the administration of different doses of an antineoplastic drug over DLD-1 cells. Under the omics paradigm, proteomics results are used to confirm the findings of the experiments.

LOX-1 and cancer: an indissoluble liaison.

Recently, a strong correlation between metabolic disorders, tumor onset, and progression has been demonstrated, directing new therapeutic strategies on metabolic targets. OLR1 gene encodes the LOX-1 receptor protein, responsible for the recognition, binding, and internalization of ox-LDL. In the past, several studied, aimed to clarify the role of LOX-1 receptor in atherosclerosis, shed light on its role in the stimulation of the expression of adhesion molecules, pro-inflammatory signaling pathways, and pro-angiogenic proteins, including NF-kB and VEGF, in vascular endothelial cells and macrophages. In recent years, LOX-1 upregulation in different tumors evidenced its involvement in cancer onset, progression and metastasis. In this review, we outline the role of LOX-1 in tumor spreading and metastasis, evidencing its function in VEGF induction, HIF-1alpha activation, and MMP-9/MMP-2 expression, pushing up the neoangiogenic and the epithelial-mesenchymal transition process in glioblastoma, osteosarcoma prostate, colon, breast, lung, and pancreatic tumors. Moreover, our studies contributed to evidence its role in interacting with WNT/APC/ß-catenin axis, highlighting new pathways in sporadic colon cancer onset. The application of volatilome analysis in high expressing LOX-1 tumor-bearing mice correlates with the tumor evolution, suggesting a closed link between LOX-1 upregulation and metabolic changes in individual volatile compounds and thus providing a viable method for a simple, non-invasive alternative monitoring of tumor progression. These findings underline the role of LOX-1 as regulator of tumor progression, migration, invasion, metastasis formation, and tumor-related neo-angiogenesis, proposing this receptor as a promising therapeutic target and thus enhancing current antineoplastic strategies.

AFM nano-mechanical study of the beating profile of hiPSC-derived cardiomyocytes beating bodies WT and DM1.

Myotonic Dystrophy type 1 (DM1) is the most common form of muscular dystrophy in adults, characterized by a variety of multisystemic features and associated with cardiac anomalies. Among cardiac phenomena, conduction defects, ventricular arrhythmias, and dilated cardiomyopathy represent the main cause of sudden death in DM1 patients. Patient-specific induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) represent a powerful in vitro model for molecular, biochemical, and physiological studies of disease in the target cells. Here, we used an Atomic Force Microscope (AFM) to measure the beating profiles of a large number of cells, organized in CM clusters (Beating Bodies, BBs), obtained from wild type (WT) and DM1 patients. We monitored the evolution over time of the frequency and intensity of the beating. We determined the variations between different BBs and over various areas of a single BB, caused by morphological and biomechanical variations. We exploited the AFM tip to apply a controlled force over the BBs, to carefully assess the biomechanical reaction of the different cell clusters over time, both in terms of beating frequency and intensity. Our measurements demonstrated differences between the WT and DM1 clusters highlighting, for the DM1 samples, an instability which was not observed in WT cells. We measured differences in the cellular response to the applied mechanical stimulus in terms of beating synchronicity over time and cell tenacity, which are in good agreement with the cellular behavior in vivo. Overall, the combination of hiPSC-CMs with AFM characterization can become a new tool to study the collective movements of cell clusters in different conditions and can be extended to the characterization of the BB response to chemical and pharmacological stimuli.

Decomposition of two-dimensional microlaser patterns.

For an ordinary individually addressable microlaser array, a separate control line is used for each microlaser, which requires a large number of control lines for even a small array. An organization that reduces the width of the control stream and simplifies packaging is matrix addressing, in which microlasers are arranged at the crossings of horizontal and vertical control lines. We consider the problem of decomposing arbitrary two-dimensional microlaser patterns into matrix-addressable patterns that are applied time sequentially to realize the target pattern. We present a mathematical model for the decomposition process and present an algorithm for optimal decomposition. We also consider bake factor, in which no more than N microlasers in a neighborhood of M (where N < M) are enabled, which avoids thermal overload by limiting the density of enabled microlasers. We conclude with a case study and show that, for completely arbitrary two-dimensional patterns, the average number of time-sequential patterns is less than the number of rows in a square array.

Optical digital processor using arrays of symmetric self-electrooptic effect devices.

Four arrays of thirty-two GaAs symmetric self-electrooptic effect devices were optically interconnected to form a looped-pipeline optical digital processor. Several circuits were demonstrated, including two shift registers and a decoder circuit. Clock frequencies of up to 1 MHz were attained. Possible extensions to and limitations of this system are described.

Connection routing for microoptic systems.

A recognized model for an all-optical digital computer consists of arrays of optical logic devices interconnected in free space with bulk optical components. A problem with this approach is that device arrays must be spaced to allow for components placed between them such as lenses, gratings, and beam splitters. The latency introduced by this spacing may be greater than device switching times, which means that tight loop processing of digital information is not possible. A solution to this problem is to replace large optical components with monolithically fabricated devices, lenses, mirrors, beam splitters, and combiners. Some connection freedom is lost due to practical limits on configurations of small components. These limits and a method to minimize their effects are explored here. It is concluded that log(2)N optical interconnects such as perfect shuffles and crossovers are not necessary for efficient digital architectures and that simple split, shift, and combine operations may be preferred for simpler optical implementations.

Design for an optical random access memory.

Cascadable optically nonlinear arrays of logic devices interconnected with space invariant optical components are proposed for the core memory of a digital computer. Access time to any portion of the memory is O(log(2)N) gate delays for logic devices with fan-in and fan-out of two, where N is the size of the memory in bits. The cost of the design in switching components is near minimal for a random access memory (RAM) between one and two components per stored bit of information depending on the size of the memory. The design is extensible to very large RAMs, although parallel access memory is preferred to a RAM configuration for large memories due to the parallel access capability of the optical design.

Optical design of a digital switch.

An optical design of a time multiplexed nonblocking space switch is described for optically nonlinear arrays of logic devices interconnected in free space. Regular interconnects in the form of crossovers are used for interconnecting optical logic devices, and some efficiency is lost due to the strict use of regular interconnects. It is shown that for this application, the higher component count is comparable with the component count for the electronic implementation of the switch. We maintain that for a high bandwidth application such as packet switching, simple bulk optics provides a suitable medium for interconnecting optical logic devices even at high speeds, and that a more elaborate approach is not warranted.

Optical design of programmable logic arrays.

Regular free-space interconnects such as the perfect shuffle and banyan provided by beam splitters, lenses, and mirrors connect optical logic gates arranged in 2-D arrays. An algorithmic design technique transforms arbitrary logic equations into a near-optimal depth circuit. Analysis shows that an arbitrary interconnect makes little or no improvement in circuit depth and can even reduce throughput. Gate count is normally higher with a regular interconnect, and we show cost bounds. We conclude that regularly interconnected circuits will have a higher gate count compared with arbitrarily interconnected circuits using the design techniques presented here and that regular free-space interconnects are comparable with arbitrary interconnects in terms of circuit depth and are preferred to arbitrary interconnects for maximizing throughput.

Crossover networks and their optical implementation.

Crossover networks are introduced as a new type of interconnection network for applications in optical computing, optical switching, and signal processing. Crossover networks belong to the class of multistage interconnection network. Two variations are presented, the half-crossover network and the full crossover network. An optical system which implements both networks is proposed and demonstrated. Crossover networks can be implemented using the full space-bandwidth product of the optical system with minimal loss of light. It is shown that crossover networks are isomorphic to other multistage networks such as the Banyan and perfect shuffle.

Digital optical computing with one-rule cellular automata.

Symbolic substitution is a parallel technique for applying pattern transformation rules to an array of ON/OFF cells. A single pattern transformation rule that requires a small fan-in and fan-out is shown to be sufficient for all computing. A simple optical architecture that makes use of optically nonlinear arrays and spaceinvariant imaging can implement the rule. Configurations of cells are arranged to realize dual-rail logic operators and connection primitives. Circuits are created by customizing a fixed 2-D mask to image light onto selected cells of the array. The presented techniques are intended to provide a foundation for work on digital optical computers.