Active and machine learning-based approaches to rapidly enhance microbial chemical production

In order to make renewable fuels and chemicals from microbes, new methods are required to engineer microbes more intelligently. Computational approaches, to engineer strains for enhanced chemical production typically rely on detailed mechanistic models (e.g., kinetic/stoichiometric models of metabolism)—requiring many experimental datasets for their parameterization—while experimental methods may require screening large mutant libraries to explore the design space for the few mutants with desired behaviors. To address these limitations, we developed an active and machine learning approach (ActiveOpt) to intelligently guide experiments to arrive at an optimal phenotype with minimal measured datasets. ActiveOpt was applied to two separate case studies to evaluate its potential to increase valine yields and neurosporene productivity in Escherichia coli. In both the cases, ActiveOpt identified the best performing strain in fewer experiments than the case studies used. This work demonstrates that machine and active learning approaches have the potential to greatly facilitate metabolic engineering efforts to rapidly achieve its objectives.     

The oviposition behavior ofAphidius ervi andMonoctonus paulensis (Hymenoptera: Aphididae) encountering different host species (Homoptera: Aphididae) in sequential patches

Two-day-old mated females ofAphidius ervi Haliday andMonoctonus paulensis (Ashmead) were each provided with two sequential host patches. Patches were comprised of plastic petri dishes containing either 15 pea aphids,Acyrthosiphum pisum (Harris), or 15 alfalfa aphids,Macrosiphum creelii Davis. Both wasp species parasitized more hosts in patches containing pea aphids than in those containing alfalfa aphids, regardless of sequence. Females ofA. ervi also laid more eggs per aphid in patches containing pea aphids than in patches containing alfalfa aphids. When both patches contained alfalfa aphids,M. paulensis females parsitized more aphids in the second patch than in the first. Fewer alfalfa aphids were parasitized in the second patch when the first patch contained pea aphids, and fewer eggs were laid per alfalfa aphid. Parasitoid females of both species exhibited consistently higher rates of oviposition into their preferred host species and adjusted their reproductive allocation to hosts and host patches as a function of their experience in previous patches.     

The bold and the shy: individual differences in rainbow trout

Boldness and shyness were investigated as 'personality' traits in hatchery-reared rainbow trout Oncorhynchus mykiss . Bold fish spent more time in an open area and were more active than shy fish and these behaviours could be used as indicators of boldness and shyness. These differences were related to learning ability in a simple conditioning task. Bold fish learned the task more quickly than shy fish.     

Developmental Lead Exposure and Two-Way Active Avoidance Training Alter the Distribution of Protein Kinase C Activity in the Rat Hippocampus

Long-term exposure to a low level of lead is associated with learning deficits. Several types of learning have been correlated to hippocampal protein kinase C (PKC) activation. This study was designed to determine if there is a correlation between the effects of lead on hippocampal PKC activation and those on learning performance. Rats were exposed to 0.2% (w/v) lead acetate at different developmental stages including a maternally exposed group, a postweaning exposed group, and a continuously exposed group. The continuously lead exposed rats tended to avoid less frequently and not respond more frequently in two-way active avoidance training than did controls. This training process was associated with translocation of hippocampal PKC activity from cytosol to membrane. Two-way analysis of variance of data indicates that there is a significant training and lead treatment interaction in the ratio of membrane to cytosolic PKC activity (F_3,32 = 3.013; p = 0.044). The interaction is attributable to the absence of the training-induced PKC translocation in the continuously lead exposed rats. In addition, no significant changes were observed in learning performance and training-induced hippocampal PKC activation after maternal and postweaning lead exposure. Continuous and longer duration of lead exposure appears to affect the learning performance and hippocampal PKC activation. These data suggest that a change in the activation of hippocampal PKC may be involved in the lead-induced deficit in learning.     

Machine Learning Approach to Detect Cardiac Arrhythmias in ECG Signals: A Survey

Cardiac arrhythmia is a condition when the heart rate is irregular either the beat is too slow or too fast. It occurs due to improper electrical impulses that coordinates the heart beats. Sudden cardiac death may occurs due to some dangerous arrhythmias conditions. Hence the main objective of the electrocardiogram (ECG) analysis is to detect the life-threatening arrhythmias accurately for appropriate treatment in order to save life. Since the last decades, several methods were reported for automatic ECG beat classifications. In this work, we present a systematic review of the current state-of-the-art methods used to detect cardiac arrhythmia using on ECG signals. It includes the signal decomposition, feature extraction and machine learning approaches used for automatic detection and decision making process. The articles covers the pre-processing, detection of QRS complex, feature extraction and classification of ECG beats. Based on the past studies, it is understood that the automated approach using computer-aided decision making process is highly required for real-time detection of cardiac arrhythmias. The advantages and limitations of different methods are discussed and also the future scopes is highlighted in the process of effective detection of cardiac arrhythmias. This study could be beneficial for researchers to analyze the existing state-of-art techniques used in detection of arrhythmia conditions.     

Possible aversion learning in the Pacific Coast wireworm

The effects of carbon dioxide and the induction of morbidity on aversion learning in larvae of the Pacific Coast wireworm Limonius canus LeConte (Coleoptera: Elateridae) are discussed. Wireworms preconditioned by exposing them one or four times to odour of Tefluthrin 20SC and Dividend XLRTA [Syngenta Crop Protection (Canada), Inc., Canada] during the induction of temporary morbidity subsequently contact tefluthrin‐treated wheat seeds in soil bioassays for as long as naïve (i.e. not preconditioned) larvae but are repelled four to five‐fold more frequently by Dividend‐treated seeds in soil bioassays than naïve wireworms, suggesting that wireworms are capable of associating a novel odour (i.e. Dividend) with morbidity but require a minimum of 10–15 min subsequent contact time with treated seeds before being repelled. Wireworms preconditioned by exposure to peppermint odour during the induction of morbidity are not subsequently repelled by peppermint odour in soil bioassays, suggesting that wireworms are either not capable of aversion learning or that the presence of a CO₂ source and/or a suitable host plant may override a negative cue (i.e. peppermint odour). In studies conducted in the absence of soil, a host plant and CO₂ production, wireworms are repelled slightly by droplets of 1.0% but not 0.1% peppermint oil. Previous exposure to peppermint odour or contact with peppermint oil‐treated filter paper during one induction of morbidity does not increase the repellency of wireworms to 1.0% peppermint oil significantly. Repellency to 1.0% peppermint oil is almost eliminated when morbidity is induced five times in the absence of peppermint odour but is restored when peppermint odour is present during preconditioning. These findings suggest that wireworm sensitivity to repellent compounds decreases when repeatedly made moribund, although the results are not sufficient to conclude that wireworms are capable of associative learning.     

Instrumental outcome devaluation with representation-mediated conditioning

In three experiments, rats were trained to perform two instrumental behaviours (R1 and R2) in the presence of discriminative stimuli (Sd1 and Sd2, respectively) to obtain a common food outcome (O1). Acquisition of the two discriminations was followed by switching the outcome accompanying R2 performance from O1 to a new one (O2). Experiment 1 showed paired presentations of O2 with a lithium chloride (LiCl) injection resulted in a reduction in the R2 performance. In the subsequent two experiments, each Sd was paired with LiCl injection and its effects on outcome consumption and instrumental performance were investigated. A reduction in the O2 consumption subsequent to the Sd devaluation was found in Experiments 2 and 3. Experiment 3 revealed a reduced R2 performance in an extinction test, following the animals’ consummatory access to the outcomes in training context. These results demonstrate representation-mediated outcome devaluation in the course of the Sd devaluation.     

Preimaginal conditioning does not affect oviposition preference in the diamondback moth

Abstract.  1. In holometabolous insects, learning has been demonstrated in both larval and adult stages. Whether learning can be retained through metamorphosis from larva via pupa to adult has long been a subject of debate. The present study is designed to distinguish between preimaginal and imaginal conditioning in the diamondback moth, Plutella xylostella L. (Lepidoptera: Plutellidae) using oviposition preference tests on females exposed to various types of learning experiences during immature and adult stages.
2. Cohorts of test insects were reared from egg to pupa on an artificial diet, or on one of two host plants, Chinese cabbage, Brassica campestris L. ssp. pekinensis , and common cabbage, Brassica oleracea L. var. capitata . The ensuing females reared on the three kinds of food showed similar oviposition preference between the two plants. A brief experience of the less preferred host, common cabbage, by adults slightly increased their preference for this plant.
3. Cohorts of test insects were reared from egg to pupa on an artificial diet with or without the addition of a neem-based oviposition deterrent (Neemix^® 4.5). Larval feeding experience did not alter oviposition response to the deterrent. However, emergence conditioning and early adult learning, achieved through experience of a residue of the deterrent carried over from the larval food on pupal cuticle and cocoons, altered oviposition preference significantly.
4. The combined results revealed no evidence of preimaginal conditioning in this insect but a strong effect of emergence conditioning and early adult learning on oviposition preference.     

Resistance to change and vulnerability to stress: autistic‐like features of GAP43‐deficient mice

There is an urgent need for animal models of autism spectrum disorder (ASD) to understand the underlying pathology and facilitate development and testing of new treatments. The synaptic growth‐associated protein‐43 (GAP43) has recently been identified as an autism candidate gene of interest. Our previous studies show many brain abnormalities in mice lacking one allele for GAP43 [GAP43 (+/?)] that are consistent with the disordered connectivity theory of ASD. Thus, we hypothesized that GAP43 (+/?) mice would show at least some autistic‐like behaviors. We found that GAP43 (+/?) mice, relative to wild‐type (+/+) littermates, displayed resistance to change, consistent with one of the diagnostic criteria for ASD. GAP43 (+/?) mice also displayed stress‐induced behavioral withdrawal and anxiety, as seen in many autistic individuals. In addition, both GAP43 (+/?) mice and (+/+) littermates showed low social approach and lack of preference for social novelty, consistent with another diagnostic criterion for ASD. This low sociability is likely because of the mixed C57BL/6J 129S3/SvImJ background. We conclude that GAP43 deficiency leads to the development of a subset of autistic‐like behaviors. As these behaviors occur in a mouse that displays disordered connectivity, we propose that future anatomical and functional studies in this mouse may help uncover underlying mechanisms for these specific behaviors. Strain‐specific low sociability may be advantageous in these studies, creating a more autistic‐like environment for study of the GAP43‐mediated deficits of resistance to change and vulnerability to stress.     

Extended Evolutionary Fast Learn-to-Walk Approach for Four-Legged Robots

Robot locomotion is an active research area. In this paper we focus on the locomotion of quadruped robots. An effective walking gait of quadruped robots is mainly concerned with two key aspects, namely speed and stability. The large search space of potential parameter settings for leg joints means that hand tuning is not feasible in general. As a result walking parameters are typically determined using machine learning techniques. A major shortcoming of using machine learning techniques is the significant wear and tear of robots since many parameter combinations need to be evaluated before an optimal solution is found.This paper proposes a direct walking gait learning approach, which is specifically designed to reduce wear and tear of robot motors, joints and other hardware. In essence we provide an effective learning mechanism that leads to a solution in a faster convergence time than previous algorithms. The results demonstrate that the new learning algorithm obtains a faster convergence to the best solutions in a short run. This approach is significant in obtaining faster walking gaits which will be useful for a wide range of applications where speed and stability are important. Future work will extend our methods so that the faster convergence algorithm can be applied to a two legged humanoid and lead to less wear and tear whilst still developing a fast and stable gait.