Materials discovery with extreme properties via reinforcement learning-guided combinatorial chemistry.

The goal of most materials discovery is to discover materials that are superior to those currently known. Fundamentally, this is close to extrapolation, which is a weak point for most machine learning models that learn the probability distribution of data. Herein, we develop reinforcement learning-guided combinatorial chemistry, which is a rule-based molecular designer driven by trained policy for selecting subsequent molecular fragments to get a target molecule. Since our model has the potential to generate all possible molecular structures that can be obtained from combinations of molecular fragments, unknown molecules with superior properties can be discovered. We theoretically and empirically demonstrate that our model is more suitable for discovering better compounds than probability distribution-learning models. In an experiment aimed at discovering molecules that hit seven extreme target properties, our model discovered 1315 of all target-hitting molecules and 7629 of five target-hitting molecules out of 100 000 trials, whereas the probability distribution-learning models failed. Moreover, it has been confirmed that every molecule generated under the binding rules of molecular fragments is 100% chemically valid. To illustrate the performance in actual problems, we also demonstrate that our models work well on two practical applications: discovering protein docking molecules and HIV inhibitors.

Prevalence and prognosis of acute pancreatitis in critically ill patients with COVID-19.

BACKGROUND: Coronavirus disease 2019 (COVID-19) is a global pandemic issue. In addition to the well-known respiratory and fever symptoms, gastrointestinal symptoms have also been reported. This study aimed to evaluate the prevalence and prognosis of patients with COVID-19 infection complicated with acute pancreatitis in intensive care unit (ICU). METHODS: This was a retrospective observational cohort study, and patients aged 18 years or older, admitted into the ICU in a single tertiary center from January 1, 2020, to April 30, 2022 were enrolled. Patients were identified by electronic medical records and reviewed manually. The primary outcome was the prevalence of acute pancreatitis among ICU patients with COVID-19. The secondary outcomes were the length of hospital stay, need for mechanical ventilation (MV), need for continuous renal replacement therapy (CRRT), and in-hospital mortality. RESULTS: A total of 4133 patients, admitted into the ICU, were screened. Among these patients, 389 were infected by COVID-19, and 86 were diagnosed with acute pancreatitis. COVID-19 positive patients were more likely to present with acute pancreatitis than COVID-19 negative patients (odds ratio = 5.42, 95% confidence interval: 2.35-6.58, P < 0.01). However, the length of hospital stay, need for MV, need for CRRT, and in-hospital mortality were not significantly different between acute pancreatitis patients with and without COVID-19 infection. CONCLUSIONS: Severe COVID-19 infections may cause acute pancreas damage in critically ill patients. However, the prognosis may not differ between acute pancreatitis patients with and without COVID-19 infection.

A new wide-angle arthroscopic system: a comparative study with a conventional 30° arthroscopic system.

PURPOSE: To compare users' hand movements in performing validated shoulder arthroscopic tasks between a 30° and a wide-angle arthroscopic system, using phantom models with an optical motion analysis system. METHODS: Twelve orthopaedic residents were enrolled and randomly allocated into two groups. In order to compensate for any learning effect, a Latin square counterbalancing technique was used. An optical motion analysis system was used with markers affixed to pre-designed sites; each participant conducted four validated shoulder arthroscopic tasks using both arthroscopic systems. Each participant was instructed to perform the experiment three times with each arthroscope. The time taken, total path length, number of movements, and average acceleration were analysed. RESULTS: Significant differences were observed for the time taken, number of movements, and average acceleration between the two arthroscopic systems (P < 0.05 for all). However, the time taken was not significant. The mean total path length measured 53 ± 38 cm with the 30° arthroscope, while the mean with the wide-angle arthroscope was significantly shorter, at 36 ± 22 cm. The mean number of movements with the 30° and wide-angle arthroscopes were 1974 ± 1305 and 1233 ± 990, respectively, while the average accelerations were 2.6 ± 1.3 and 1.2 ± 0.6 cm/s(2), respectively. The mean time taken was 13 % faster when using the wide-angle arthroscopic system, although this was not statistically significant. CONCLUSION: The wide-angle arthroscopic system improved the arthroscope manoeuvre in terms of the total path length, number of movements, and average acceleration required for experimental arthroscopy. This system may help surgeons triangulate the arthroscope and surgical instruments during surgery by expanding the field of view.

Negative self-regulation of TLR9 signaling by its N-terminal proteolytic cleavage product.

TLR signaling is essential to innate immunity against microbial invaders and must be tightly controlled. We have previously shown that TLR9 undergoes proteolytic cleavage processing by lysosomal proteases to generate two distinct fragments. The C-terminal cleavage product plays a critical role in activating TLR9 signaling; however, the precise role of the N-terminal fragment, which remains in lysosomes, in the TLR9 response is still unclear. In this article, we report that the N-terminal cleavage product negatively regulates TLR9 signaling. Notably, the N-terminal fragment promotes the aspartic protease-mediated degradation of the C-terminal fragment in endolysosomes. Furthermore, the N-terminal TLR9 fragment physically interacts with the C-terminal product, thereby inhibiting the formation of homodimers of the C-terminal fragment; this suggests that the monomeric C-terminal product is more susceptible to attack by aspartic proteases. Together, these results suggest that the N-terminal TLR9 proteolytic cleavage product is a negative self-regulator that prevents excessive TLR9 signaling activity.