Stefano Vassanelli

Rapid advances in hardware-based technologies during the past decades have opened up new possibilities for life scientists to gather multimodal data in various application domains, such as omics, bioimaging, medical imaging, and (brain/body)-machine interfaces. These have generated novel opportunities for development of dedicated data-intensive machine learning techniques. In particular, recent research in deep learning (DL), reinforcement learning (RL), and their combination (deep RL) promise to revolutionize the future of artificial intelligence. The growth in computational power accompanied by faster and increased data storage, and declining computing costs have already allowed scientists in various fields to apply these techniques on data sets that were previously intractable owing to their size and complexity. This paper provides a comprehensive survey on the application of DL, RL, and deep RL techniques in mining biological data. In addition, we compare the performances of DL techniques when applied to different data sets across various application domains. Finally, we outline open issues in this challenging research area and discuss future development perspectives.

We have studied the induction of the mitochondrial cyclosporin A-sensitive permeability transition pore (PTP) by the bifunctional SH group reagent phenylarsine oxide (PhAsO). Addition of nanomolar concentrations of the electroneutral H(+)-K+ ionophore nigericin to nonrespiring mitochondria in sucrose medium determines a dramatic increase of the time required for PTP induction by PhAsO, while no effect of nigericin is apparent in KCl medium. Using mitochondria loaded with the internal pH indicator 2',7'-bis(carboxyethyl)-5(6)-carboxyfluorescein, we show that the effect of nigericin is mediated by the ionophore-induced acidification of matrix pH. Indeed, experimental manipulation of pHi by a number of treatments indicates that PTP induction is directly related to matrix pH, in that the PTP induction process becomes slower as pHi decreases at constant pHo. PTP induction by PhAsO in respiration-inhibited mitochondria is stimulated by Ca2+ and inhibited by a series of divalent cations. Since PhAsO induces the PTP even in the presence of excess EGTA and in the absence of respiration (Lenartowicz, E., Bernardi, P., and Azzone, G.F. (1991) J. Bioenerg. Biomembr. 23, 679-688), we have been able to study the Ca2+ dependence of the induction process. We show that the apparent Km for Ca2+ activation is about 10(-5) M and that Ca2+, cyclosporin A, and inhibitory Me2+ ions behave as if they were competing for the same binding site(s) on the pore. Since similar results are obtained from patch-clamp experiments on the mitochondrial megachannel (Szabó, I., Bernardi, P., and Zoratti, M. (1992) J. Biol. Chem. 267, 2940-2946), we suggest that (i) the PTP and the mitochondrial megachannel are the same molecular structures and (ii) the same factors affect both the process of pore induction and its open-closed orientation.

Uncoupling protein mediates electrophoretic transport of protons and anions across the inner membrane of brown adipose tissue mitochondria. The mechanism and site of proton transport, the mechanism by which fatty acids activate proton transport, and the relationship between fatty acids and anion transport are unknown. We used fluorescent probes to measure H+ and anion transport in vesicles reconstituted with purified uncoupling protein and carried out a comparative study of the effects of laurate and its close analogue, undecanesulfonate. Undecanesulfonate was transported by uncoupling protein with a Km value similar to that observed for laurate as it activated H+ transport. Both laurate and undecanesulfonate inhibited Cl- with competitive kinetics. Undecanesulfonate inhibited laurate-induced H+ transport with competitive kinetics. Undecanesulfonate and laurate differed in two important respects. (i) Laurate caused uncoupling protein-mediated H+ transport, whereas undecanesulfonate did not. (ii) Lauric acid was rapidly transported across the bilayer by nonionic diffusion, whereas undecanesulfonic was not. We infer that the role of uncoupling protein in H+ transport is to transport fatty acid anions and that fatty acids induce H+ transport because they can diffuse electroneutrally across the membrane. According to this hypothesis, uncoupling protein is a pure anion porter and does not transport protons; rather it is designed to enable fatty acids to behave as cycling protonophores.

Advanced brain-chip interfaces with numerous recording sites bear great potential for investigation of neuroprosthetic applications. The bottleneck towards achieving an efficient bio-electronic link is the real-time processing of neuronal signals, which imposes excessive requirements on bandwidth, energy and computation capacity. Here we present a unique concept where the intrinsic properties of memristive devices are exploited to compress information on neural spikes in real-time. We demonstrate that the inherent voltage thresholds of metal-oxide memristors can be used for discriminating recorded spiking events from background activity and without resorting to computationally heavy off-line processing. We prove that information on spike amplitude and frequency can be transduced and stored in single devices as non-volatile resistive state transitions. Finally, we show that a memristive device array allows for efficient data compression of signals recorded by a multi-electrode array, demonstrating the technology's potential for building scalable, yet energy-efficient on-node processors for brain-chip interfaces.