Comparing storm resolving models and climates via unsupervised machine learning.

Global storm-resolving models (GSRMs) have gained widespread interest because of the unprecedented detail with which they resolve the global climate. However, it remains difficult to quantify objective differences in how GSRMs resolve complex atmospheric formations. This lack of comprehensive tools for comparing model similarities is a problem in many disparate fields that involve simulation tools for complex data. To address this challenge we develop methods to estimate distributional distances based on both nonlinear dimensionality reduction and vector quantization. Our approach automatically learns physically meaningful notions of similarity from low-dimensional latent data representations that the different models produce. This enables an intercomparison of nine GSRMs based on their high-dimensional simulation data (2D vertical velocity snapshots) and reveals that only six are similar in their representation of atmospheric dynamics. Furthermore, we uncover signatures of the convective response to global warming in a fully unsupervised way. Our study provides a path toward evaluating future high-resolution simulation data more objectively.

Improving product quality and productivity of bispecific molecules through the application of continuous perfusion principles.

Bispecific protein scaffolds can be more complex than traditional monoclonal antibodies (MAbs) because two different sites/domains for epitope binding are needed. Because of this increased molecular complexity, bispecific molecules are difficult to express and can be more prone to physical and chemical degradation compared to MAbs, leading to higher levels of protein aggregates, clipped species, or modified residues in cell culture. In this study, we investigated cell culture performance for the production of three types of bispecific molecules developed at Amgen. In particular, we cultured a total of six CHO cell lines in both an approximately 12-day fed-batch process and an approximately 40-day high-density perfusion process. Harvested cell culture fluid from each process was purified and analyzed for product quality attributes including aggregate levels, clipped species, charge variants, individual amino acid modifications and host cell protein (HCP) content. Our studies showed that in average, the intensified perfusion process increased 15-fold the integrated viable cell density and the total harvested product (and fivefold the daily volumetric productivity) compared to fed-batch. Furthermore, bispecific product quality improved in perfusion culture (as analyzed in affinity-capture pools) with reduction in levels of aggregates (up to 72% decrease), clipped species (up to 75% decrease), acidic variants (up to 76% decrease), deamidated/isomerized species in complementarity-determining regions, and HCP (up to 84% decrease). In summary, the intensified perfusion process exhibited better productivity and product quality, highlighting the potential to use it as part of a continuous manufacturing process for bispecific scaffolds.