Search for: All records

In human pedagogy, teachers and students can interact adaptively to maximize communication efficiency. The teacher adjusts her teaching method for different students, and the student, after getting familiar with the teacher’s instruction mechanism, can infer the teacher’s intention to learn faster. Recently, the benefits of integrating this cooperative pedagogy into machine concept learning in discrete spaces have been proved by multiple works. However, how cooperative pedagogy can facilitate machine parameter learning hasn’t been thoroughly studied. In this paper, we propose a gradient optimization based teacher-aware learner who can incorporate teacher’s cooperative intention into the likelihood function and learn provably faster compared with the naive learning algorithms used in previous machine teaching works. We give theoretical proof that the iterative teacher-aware learning (ITAL) process leads to local and global improvements. We then validate our algorithms with extensive experiments on various tasks including regression, classification, and inverse reinforcement learning using synthetic and real data. We also show the advantage of modeling teacher-awareness when agents are learning from human teachers. 

As upstream product titers increase, the downstream chromatographic capture step has become a significant “downstream bottleneck.” Precipitation becomes more attractive under these conditions as the supersaturation driving force increases with the ever-increasing titer. In this study, two precipitating reagents with orthogonal mechanisms, polyethylene glycol (PEG) as a volume excluder and zinc chloride (ZnCl2) as a cross linker, were examined as precipitants for two monoclonal antibodies (mAbs), one stable and the other aggregation-prone, in purified drug substance and harvested cell culture fluid forms. Manual batch solubility and redissolution experiments were performed as scouting experiments. A high throughput (HTP) liquid handling system was used to investigate the design space as fully as possible while reducing time, labor, and material requirements. Precipitation and redissolution were studied by systematically varying the concentrations of PEG and ZnCl2 to identify combinations that resulted in high yield and good quality for the stable mAb; PEG concentrations in the range 7–7.5 wt/vol% together with 10 mM ZnCl2 gave a yield of 97% and monomer contents of about 93%. While yield for the unstable mAb was high, quality was not acceptable. Performance at selected conditions was further corroborated for the stable mAb using a continuous tubular precipitation reactor at the laboratory scale. The HTP automation system was a powerful tool for locating desired (customized) conditions for antibodies of different physicochemical properties. 

There is renewed interest in the possibility of using precipitation for initial capture of high-value therapeutic proteins as part of an integrated continuous downstream process. Precipitation is greatly facilitated by the high product titers now achieved in most cell culture processes, in sharp contrast to chromatographic processes whose performance is reduced at high titers. The current study used a combination of reversible cross-linking (zinc chloride, ZnCl2) and volume exclusion (polyethylene glycol) agents to precipitate a monoclonal antibody product directly from harvested cell culture fluid using a continuous tubular precipitation reactor. The precipitates were then dewatered and continuously washed using tangential flow filtration, with a countercurrent-staged configuration used to reduce the amount of wash buffer required and increase host cell protein removal. Long-term operation was achieved by operating the membrane modules below the critical filtrate flux to avoid fouling. Experimental results demonstrate the feasibility of this fully continuous integrated precipitation process at bench scale, with design calculations used to explore the key factors affecting the performance of this system for initial antibody capture. 

Abstract Disease modelling has had considerable policy impact during the ongoing COVID-19 pandemic, and it is increasingly acknowledged that combining multiple models can improve the reliability of outputs. Here we report insights from ten weeks of collaborative short-term forecasting of COVID-19 in Germany and Poland (12 October–19 December 2020). The study period covers the onset of the second wave in both countries, with tightening non-pharmaceutical interventions (NPIs) and subsequently a decay (Poland) or plateau and renewed increase (Germany) in reported cases. Thirteen independent teams provided probabilistic real-time forecasts of COVID-19 cases and deaths. These were reported for lead times of one to four weeks, with evaluation focused on one- and two-week horizons, which are less affected by changing NPIs. Heterogeneity between forecasts was considerable both in terms of point predictions and forecast spread. Ensemble forecasts showed good relative performance, in particular in terms of coverage, but did not clearly dominate single-model predictions. The study was preregistered and will be followed up in future phases of the pandemic.