Search for: All records

The C‐terminal decarboxylation of peptides provides an important opportunity to synthesize modern peptide pharmaceuticals that contain C‐terminal amides. This transformation can be achieved by electrochemical oxidation; however, the standard implementation depends on oxidation potential for selectivity which may represent a challenge when amino acid residues containing electroactive side chains are present. To address this limitation, an alternative mechanistic paradigm has been introduced for selective decarboxylation of a C‐terminal carboxylate, one that relies on a chelation event. In a proof‐of‐principle experiment used to probe and define the viability of this mechanism, it is demonstrated that the combination of an iron mediator and a C‐terminal glutamate residue can be used to conduct the reaction in the presence of the more electron‐rich tyrosine residue frequently found in medicinally active peptides. Investigations into the reaction specifics and the scope/limitations provide key insights into the reaction mechanism and how such processes can be optimized. The success of the method highlighted here points to a more general binding‐based approach to drive C‐terminal decarboxylation that utilizes a functional group motif not possible at any other position in a peptide. 

At California State Polytechnic University, Pomona (Cal Poly Pomona, CPP), two gatekeeper courses for the undergraduate students in the Civil Engineering program have been identified as Statics and Mechanics of Materials. Our university’s Civil Engineering Department is the largest undergraduate CE Department in the nation with approximately 1,600 students, graduating 15% of California’s Civil Engineers. Identifying sources of students’ struggles and proposing effective interventions to support students’ success is crucial. As these two gatekeeper courses serve as prerequisites to many engineering courses, low performance in these courses contributes to a large dropout rate, delayed graduation, and continued poor performance in subsequent courses. To understand students’ struggles, historical data between Fall 2018 to Spring 2022 was examined, including: (a) failure rates for the gatekeeper courses, (b) achievement gaps (the difference between under-represented minority students (URM) and non-URM students), and (c) the correlation between students’ grades in the gatekeeper courses compared to their upper division engineering courses. A comprehensive literature review was conducted to identify proven best practices for improving student performance in STEM disciplines. The literature highlights the effectiveness of targeted interventions, as follows: (1) prepare all students for success in the gatekeeper courses and close the achievement gaps, through a Summer Bridge Program, (2) improve the students’ performance in Statics, Mechanics of Materials, and subsequent courses, and reduce Time-to-Degree, and (3) address variability in teaching between all instructors through training workshops. This paper provides a review of interventions utilized to write a proposal to request funding to agencies such as National Science Foundation and offers actionable insights for educators seeking to reduce failure rates, close achievement gaps, and standardize instructional quality across courses. 

Plant phytochromes are well-studied photoreceptors that sense red and far-red light, regulating photomorpho- genic development. Molecular signaling mechanisms of phytochrome A (phyA) and phyB largely overlap, especially in regulation of PHYTOCHROME-INTERACTING FACTORs (PIFs) and E3 ligase complexes composed of CONSTITUTIVELY PHOTOMORPHOGENIC 1 (COP1) and SUPPRESSORs OF phyA-105 (SPAs). However, the differences in their molecular signaling mechanisms remain unclear. Constitutively active mutants of phyB (YVB) and NLS-fused phyA (YVA:NLS) mediate light-independent seedling development, leading to constitutive photomorphogenic (cop) phenotypes in their transgenic Arabidopsis plants. Interestingly, YVB interacted with PIF3 independently of light, but YVA showed little interaction. In this study, we investigated distinct signaling mechanisms underlying the similar cop phenotypes given by YVB and YVA:NLS. Our findings indicated that YVA efficiently inactivate the COP1/SPA complex, leading to accumulation of ELONGATED HYPOCOTYL 5 (HY5) and subsequent expression of its target genes HY5 and HYH. YVB induced light-independent PIF3 and PIF1 degra- dation, in addition to HY5 accumulation. Moreover, co-expression of PIF3 in the YVB plant significantly attenuated the cop phenotypes, but minimal effects were observed in the YVA:NLS plant. In particular, PIF3 negatively regulated the interaction between YVB and COP1, which decreased HY5 accumulation in the YVB plant co-expressing PIF3. Furthermore, when transferred from light to dark, PIF3 was highly accumulated in phyB-5, whereas HY5 is degraded faster in phyA-201 compared to that in Ler. Collectively, our results suggest HY5 accumulation as the molecular bases for the cop phenotypes and also indicate that phyB is more important for regulating PIF3, whereas phyA effectively inactivates the COP1/SPA complex relative to PIF3 degradation. 

Submitted Article 

We introduce randomized algorithms to Clifford's Geometric Algebra, generalizing randomized linear algebra to hypercomplex vector spaces. This novel approach has many implications in machine learning, including training neural networks to global optimality via convex optimization. Additionally, we consider fine-tuning large language model (LLM) embeddings as a key application area, exploring the intersection of geometric algebra and modern AI techniques. In particular, we conduct a comparative analysis of the robustness of transfer learning via embeddings, such as OpenAI GPT models and BERT, using traditional methods versus our novel approach based on convex optimization. We test our convex optimization transfer learning method across a variety of case studies, employing different embeddings (GPT-4 and BERT embeddings) and different text classification datasets (IMDb, Amazon Polarity Dataset, and GLUE) with a range of hyperparameter settings. Our results demonstrate that convex optimization and geometric algebra not only enhances the performance of LLMs but also offers a more stable and reliable method of transfer learning via embeddings.