Enzymatic activity of fibroblast activation protein-α is essential for TGF-ß1-induced fibroblastic differentiation of human periodontal ligament cells.

Human periodontal ligament cells (hPDLCs) contain multipotent postnatal stem cells that can differentiate into PDL fibroblasts, osteoblasts, and cementoblasts. Interaction between the extracellular environment and stem cells is an important factor for differentiation into other progenitor cells. To identify cell surface molecules that induce PDL fibroblastic differentiation, we developed a series of monoclonal antibodies against membrane/ECM molecules. One of these antibodies, an anti-PDL25 antibody, recognizes approximately a 100 kDa protein, and this antigenic molecule accumulates in the periodontal ligament region of tooth roots. By mass spectrometric analysis, we found that the antigenic molecule recognized by the anti-PDL25 antibody is fibroblast activation protein α (FAPα). The expression level of FAPα/PDL25 increased in TGF-ß1-induced PDL fibroblasts, and this protein was localized in the cell boundaries and elongated processes of the fibroblastic cells. Ectopic expression of FAPα induced fibroblastic differentiation. In contrast, expression of representative markers for PDL differentiation was decreased by knock down and antibody blocking of FAPα/PDL25. Inhibition of dipeptidyl peptidase activity by a potent FAPα inhibitor dramatically inhibited PDL fibroblastic marker expression but did not affect in cell proliferation and migration.

Synthesis, nonlinear optical activity, solvents effect, ß-cyclodextrin effect, and cytotoxic activity on skin fibroblast and breast cancer cell lines of a new chalcone derivative of nabumetone.

The use of small organic molecules, such as chalcones, for efficient applications as organic luminescent materials has attracted increasing attention owing to their interesting optical, photophysical, and biological properties. In this study, a new chalcone, 1-(4-isopropylphenyl)-5-(6-methoxynaphthalen-2-yl)pent-1-en-3-one (INM), was synthesized via base condensation between nabumetone and cuminaldehyde. INM was subsequently identified and characterized by FT-IR, NMR spectroscopy (1H and 13C), mass spectrometry, elemental analysis, X-ray diffraction, thermogravimetric analysis, and FESEM studies. Investigation of the solvent effect revealed that the π â†’ π* transition involved a bathochromic shift from hexane to water and a large Stokes-shifted, twisted intramolecular charge-transfer emission in water. Diffuse reflectance spectral studies confirmed the formation of transparent INM chalcones with excellent crystallinity, and photoluminescence studies substantiated the low recombination rate of electrons and holes. Tauc plot analysis with the Kubelka-Munk algorithm revealed higher direct (3.57 eV) and indirect (3.41 eV) bandgap energies of INM. Density functional theory calculations at B3LYP/6-31G(d,p) revealed that INM had promising nonlinear optical activity (ß ≈ 30.504 × 10-30 compared to a reference material, urea. Cell biocompatibility was evaluated after culturing skin fibroblasts and breast cancer cells with INM using the MTT assay and fluorescence microscopy of the live/dead cell assay. It was observed that INM exhibited good NIH/3T3 cell adhesion and proliferation and the weak inhibiting ability of MDA-MB231.

Benchtop IR Imaging of Live Cells: Monitoring the Total Mass of Biomolecules in Single Cells.

Absolute quantity imaging of biomolecules on a single cell level is critical for measurement assurance in biosciences and bioindustries. While infrared (IR) transmission microscopy is a powerful label-free imaging modality capable of chemical quantification, its applicability to hydrated biological samples remains challenging due to the strong IR absorption by water. Traditional IR imaging of hydrated cells relies on powerful light sources, such as synchrotrons, to mitigate the light absorption by water. However, we overcome this challenge by applying a solvent absorption compensation (SAC) technique to a home-built benchtop IR microscope based on an external-cavity quantum cascade laser. SAC-IR microscopy adjusts the incident light using a pair of polarizers to precompensate the IR absorption by water while retaining the full dynamic range. Integrating the IR absorbance over a cell yields the total mass of biomolecules per cell. We monitor the total mass of the biomolecules of live fibroblast cells over 12 h, demonstrating promise for advancing our understanding of the biomolecular processes occurring in live cells on the single-cell level.

Configurable Synaptic and Stochastic Neuronal Functions in ZnTe-Based Memristor for an RBM Neural Network.

This study presents findings that demonstrate the possibility of simplifying neural networks by inducing multifunctionality through separate manipulation within a single material. Herein, two-terminal memristor W/ZnTe/W devices implemented a multifunctional memristor comprising a selector, synapse, and a neuron using an ovonic threshold switching material. By setting the low-current level (µA) in the forming process, a stable memory-switching operation is achieved, and the capacity to implement a synapse is demonstrated based on paired-pulse facilitation/depression, potentiation/depression, spike-amplitude-dependent plasticity, and spike-number-dependent plasticity outcomes. Based on synaptic behavior, the Modified National Institute of Standards and Technology database image classification accuracy is up to 90%. Conversely, by setting the high-current level (mA) in the forming process, the stable bipolar threshold switching operation and good selector characteristics (300 ns switching speed, free-drift, recovery properties) are demonstrated. In addition, a stochastic neuron is implemented using the stochastic switching response in the positive voltage region. Utilizing stochastic neurons, it is possible to create a generative restricted Boltzmann machine model.

In vitro-irradiated cancer vaccine enhances anti-tumor efficacy of radiotherapy and PD-L1 blockade in a syngeneic murine breast cancer model.

BACKGROUND AND PURPOSE: Local radiotherapy (RT) exerts immunostimulatory effects by inducing immunogenic cell death. However, it remains unknown whether in vitro-irradiated tumor cells can elicit anti-tumor responses and enhance the efficacy of local RT and immune checkpoint inhibitors when injected in vivo. METHODS AND MATERIALS: We tested the "in vitro-irradiated cancer vaccine (ICV)", wherein tumor cells killed by varying doses of irradiation and their supernatants are intravenously injected. We examined the efficacy of combining local RT (24 Gy in three fractions), PD-L1 blockade, and the ICV in a murine breast cancer model. The immune cell profiles were analyzed via flow cytometry and immunohistochemistry. The cytokine levels were measured by multiplex immunoassays. RESULTS: The ICV significantly increased the effector memory phenotype and interferon-γ production capacity in splenic CD8+ T cells. The in vitro-irradiated products contained immune response-related molecules. When combined with local RT and PD-L1 blockade, the ICV significantly delayed the growth of irradiated and non-irradiated tumors. The triple combination therapy increased the proportions of CD8+ T cells and effector memory CD8+ T cells while decreasing the proportion of CTLA-4+ exhausted CD8+ T cells within tumor microenvironment. Additionally, plasma level of interferon-γ and proliferation of effector T cells in the spleen and tumor-draining lymph nodes were significantly increased by the triple combination therapy. CONCLUSIONS: The ICV enhanced the therapeutic efficacy of local RT and PD-L1 blockade by augmenting anti-tumor immune responses. Our findings suggest a therapeutic potential of in vitro-irradiation products of tumor cells.

Charting Proficiency: The Learning Curve in Robotic Hysterectomy for Large Uteri Exceeding 1000 g.

Background/Objectives: This study evaluates the safety and surgical outcomes of performing robotic hysterectomy on uteri weighing over 1000 g, with a focus on the surgeon's learning curve. Methods: A retrospective analysis was conducted on 44 patients who underwent hysterectomy by a single surgeon from January 2020 to February 2024 using the DaVinci Xi System. Surgical procedures included total hysterectomy with bilateral salpingectomy, and specimens were removed via transvaginal manual morcellation. Operative times were segmented into docking, console, morcellation, and conversion times. Results: Results indicated an inflection point in the 20th case, suggesting proficiency after 20 surgeries. Comparison between early (Group A, cases 1-20) and later cases (Group B, cases 21-44) showed significant reductions in console time (CT) and morcellation time (MT) in Group B, leading to a shorter overall operative time (OT). Although estimated blood loss was higher in Group A, it was not statistically significant. Hemoglobin differences were significantly higher in Group B. No significant differences were observed in transfusion rates, postoperative analgesic usage, or complications between the groups. Conclusions: The study concludes that robotic hysterectomy for large uteri is safe and that surgical proficiency improves significantly after 20 cases, enhancing overall outcomes.

Learning Curve Analysis of Single-Incision Ovarian Cystectomy: Comparative Study of Robotic and Conventional Laparoscopic Techniques.

Ovarian cystectomy, aimed at preserving fertility, has advanced through minimally invasive surgical techniques. This study evaluates the learning curves and surgical outcomes of three such approaches: DaVinci Robotic Single-Site (RSS), DaVinci Robotic Single-Port (RSP), and laparo-endoscopic single-site surgery (LESS). To analyze the learning curves and surgical outcomes for these techniques, providing insights into their effectiveness and proficiency development. Retrospective analysis of 104 patients with ovarian tumors, divided into RSS (n = 52), RSP (n = 22), and LESS (n = 30) groups. Metrics analyzed included age, BMI, tumor size, hemoglobin drop, operative time, docking time, console time, and tumor location. No significant differences in age, BMI, transfusion rate, hemoglobin drop, or length of stay were found among the groups. RSS had larger tumors on average, and LESS had a higher occurrence rate on the right side. LESS demonstrated the shortest operative time, while RSS and RSP had comparable times. Docking and console times did not differ significantly between RSS and RSP. RSP reached proficiency faster than RSS in docking and console times, while LESS exhibited the greatest variability in operative time. RSP offers a faster and more consistent learning curve, making it advantageous for complex procedures, whereas LESS provides shorter operative times but with higher variability. These findings are crucial for surgical training and resource allocation in medical institutions.

Is It the Best Option? Robotic Surgery for Endometriosis.

Endometriosis is a chronic condition affecting approximately 10% of women of reproductive age, leading to significant physical and emotional stress. Treatments include medical management and surgical interventions, with laparoscopic surgery being the gold standard for removing endometrial tissue. The advent of robotic-assisted laparoscopic surgery (RALS) has enabled more complex procedures to be performed minimally invasively, increasing its use in high-difficulty surgeries. Developed in the late 20th century, systems like the Da Vinci Surgical System have revolutionized surgery by enhancing precision, dexterity, and visualization. The latest models, including the Da Vinci Xi and SP, offer advanced features such as enhanced arm mobility, fluorescence imaging, and single-port capabilities. Comparative studies of RALS and conventional laparoscopy (LPS) for endometriosis show mixed results. While some studies indicate no significant differences in complications or recovery outcomes, others highlight longer operative times and hospital stays for RALS. Despite these drawbacks, RALS is not inferior to LPS overall. The clinical benefits of RALS include greater precision and accuracy, reduced surgeon fatigue, and a faster learning curve, facilitated by advanced ergonomic and control systems. However, the high costs and extensive infrastructure requirements limit the accessibility and availability of robotic surgery, particularly in smaller or rural hospitals. The absence of tactile feedback remains a challenge, though upcoming advancements aim to address this. Continued research and development are essential to make robotic surgery more cost-effective and broadly accessible, ensuring its benefits can reach a wider patient population. This abstract encapsulates the key aspects of robotic surgery's development, comparative studies with conventional methods, and its clinical benefits and limitations, highlighting the need for ongoing improvements and research.

Quantum Cascade Laser Infrared Spectroscopy for Glycan Analysis of Glycoprotein Solutions.

Glycans are oligosaccharides attached to proteins or lipids and affect their functions, such as drug efficacy, structural contribution, metabolism, immunogenicity, and molecular recognition. Conventional glycosylation analysis has relied on destructive, slow, system-sensitive methods, including enzymatic reactions, chromatography, fluorescence labeling, and mass spectrometry. Herein, we propose quantum cascade laser (QCL) infrared (IR) spectroscopy as a rapid, nondestructive method to quantify glycans and their monosaccharide composition. Previously, we demonstrated high-sensitivity IR spectroscopy of protein solution using solvent absorption compensation (SAC) and double-beam modulation (DBM) techniques. However, the SAC-DBM approach suffered a limited frequency scanning range (<400 cm-1) due to the light dispersion by acousto-optic modulators (AOMs). Here, we implemented a mirror-based double-pass AOM in the SAC-DBM scheme and successfully extended the frequency range to (970 to 1840 cm-1), which encompasses the vibrational fingerprint of biomolecules. The extended frequency range allowed the simultaneous observation of monosaccharide ring bands (1000 to 1200 cm-1) and protein amide bands (1500 to 1700 cm-1). We compared the IR spectra of six glycoproteins and two nonglycosylated proteins with the results from intact mass spectrometry. The IR absorbance ratios of the ring band to the amide band of glycoproteins in solutions showed a linear correlation with the ratios of glycan to protein backbone masses. Furthermore, a multivariate analysis produced monosaccharide compositions consistent with the reported database for the glycoproteins, and the monosaccharide compositions were used to improve the predictability of the glycan-protein mass ratio from the IR-absorbance ratio. This nondestructive, high-sensitivity QCL-IR spectroscopy could be used as a standard method to monitor batch-to-batch comparability during drug manufacturing and quantify the glycosylation and monosaccharide composition of new glycoproteins and other glycosylated biosystems.

Large Language Models for Inorganic Synthesis Predictions.

We evaluate the effectiveness of pretrained and fine-tuned large language models (LLMs) for predicting the synthesizability of inorganic compounds and the selection of precursors needed to perform inorganic synthesis. The predictions of fine-tuned LLMs are comparable to─and sometimes better than─recent bespoke machine learning models for these tasks but require only minimal user expertise, cost, and time to develop. Therefore, this strategy can serve both as an effective and strong baseline for future machine learning studies of various chemical applications and as a practical tool for experimental chemists.

Reaction Templates: Bridging Synthesis Knowledge and Artificial Intelligence.

ConspectusThe field of chemical research boasts a long history of developing software to automate synthesis planning and reaction prediction. Early software relied heavily on expert systems, requiring significant effort to encode vast amounts of synthesis knowledge into a computer-readable format. However, recent advancements in deep learning have shifted the focus toward AI models, offering improved prediction capabilities. Despite these advancements, current AI models often lack the integration of known synthesis rules and intuitions, creating a gap that hinders interpretability and future development of the models. To bridge them, our research group has been actively working on incorporating reaction templates into deep learning models, achieving promising results across various applications.In this Account, we present our latest works to incorporate the known synthesis knowledge into the deep learning models through the utilization of reaction templates. We begin by highlighting the limitations of early computer programs heavily reliant on hand-coded rules. These programs, while providing a foundation for the field, presented limitations in scalability and adaptability. We then introduce SMARTS (SMILES arbitrary target specification), a popular Python-readable format for representing chemical reactions. This format of reaction encoding facilitates the quick integration of synthesis knowledge into AI models built using the Python language. With the SMARTS-based reaction templates, we introduce our recent efforts of developing an AI model for reaction-based molecule optimization. Subsequently, we discuss the recent efforts to automate the extraction of reaction templates from vast chemical reaction databases. This approach eliminates the previously required manual effort of encoding knowledge, a process that could be time-consuming and prone to error when dealing with large data sets. By customizing the automated extraction algorithm, we have developed powerful AI models for specific tasks such as retrosynthesis (LocalRetro), reaction outcome prediction (LocalTransform), and atom-to-atom mapping (LocalMapper). These models, aligned with the intuition of chemists, demonstrate the effectiveness of incorporating reaction templates into deep learning frameworks.Looking toward the future, we believe that utilizing reaction templates to connect known chemical knowledge and AI models holds immense potential for various applications. Not only can this approach significantly benefit future AI models focused on challenging tasks like reaction mechanism labeling and prediction, but we anticipate it can also extend its reach to the realm of inorganic synthesis. By integrating synthesis knowledge, we can not only achieve improved performance but also enhance the interpretability of AI models, paving the way for further advancements in AI-powered chemical synthesis.

On the Analysis of Inter-Relationship between Auto-Scaling Policy and QoS of FaaS Workloads.

A recent development in cloud computing has introduced serverless technology, enabling the convenient and flexible management of cloud-native applications. Typically, the Function-as-a-Service (FaaS) solutions rely on serverless backend solutions, such as Kubernetes (K8s) and Knative, to leverage the advantages of resource management for underlying containerized contexts, including auto-scaling and pod scheduling. To take the advantages, recent cloud service providers also deploy self-hosted serverless services by facilitating their on-premise hosted FaaS platforms rather than relying on commercial public cloud offerings. However, the lack of standardized guidelines on K8s abstraction to fairly schedule and allocate resources on auto-scaling configuration options for such on-premise hosting environment in serverless computing poses challenges in meeting the service level objectives (SLOs) of diverse workloads. This study fills this gap by exploring the relationship between auto-scaling behavior and the performance of FaaS workloads depending on scaling-related configurations in K8s. Based on comprehensive measurement studies, we derived the logic as to which workload should be applied and with what type of scaling configurations, such as base metric, threshold to maximize the difference in latency SLO, and number of responses. Additionally, we propose a methodology to assess the scaling efficiency of the related K8s configurations regarding the quality of service (QoS) of FaaS workloads.

Highly efficient microbial inactivation enabled by tunneling charges injected through two-dimensional electronics.

Airborne pathogens retain prolonged infectious activity once attached to the indoor environment, posing a pervasive threat to public health. Conventional air filters suffer from ineffective inactivation of the physics-separated microorganisms, and the chemical-based antimicrobial materials face challenges of poor stability/efficiency and inefficient viral inactivation. We, therefore, developed a rapid, reliable antimicrobial method against the attached indoor bacteria/viruses using a large-scale tunneling charge-motivated disinfection device fabricated by directly dispersing monolayer graphene on insulators. Free charges can be stably immobilized under the monolayer graphene through the tunneling effect. The stored charges can motivate continuous electron loss of attached microorganisms for accelerated disinfection, overcoming the diffusion limitation of chemical disinfectants. Complete (>99.99%) and broad-spectrum disinfection was achieved <1 min of attachment to the scaled-up device (25 square centimeters), reliably for 72 hours at high temperature (60°C) and humidity (90%). This method can be readily applied to high-touch surfaces in indoor environments for pathogen control.

Genetically engineered bacteriophages as novel nanomaterials: applications beyond antimicrobial agents.

Bacteriophages, also known as phages, are viruses that replicate in bacteria and archaea. Phages were initially discovered as antimicrobial agents, and they have been used as therapeutic agents for bacterial infection in a process known as "phage therapy." Recently, phages have been investigated as functional nanomaterials in a variety of areas, as they can function not only as therapeutic agents but also as biosensors and tissue regenerative materials. Phages are nontoxic to humans, and they possess self-assembled nanostructures and functional properties. Additionally, phages can be easily genetically modified to display specific peptides or to screen for functional peptides via phage display. Here, we demonstrated the application of phage nanomaterials in the context of tissue engineering, sensing, and probing.

Quantum annealing-aided design of an ultrathin-metamaterial optical diode.

Thin-film optical diodes are important elements for miniaturizing photonic systems. However, the design of optical diodes relies on empirical and heuristic approaches. This poses a significant challenge for identifying optimal structural models of optical diodes at given wavelengths. Here, we leverage a quantum annealing-enhanced active learning scheme to automatically identify optimal designs of 130 nm-thick optical diodes. An optical diode is a stratified volume diffractive film discretized into rectangular pixels, where each pixel is assigned to either a metal or dielectric. The proposed scheme identifies the optimal material states of each pixel, maximizing the quality of optical isolation at given wavelengths. Consequently, we successfully identify optimal structures at three specific wavelengths (600, 800, and 1000 nm). In the best-case scenario, when the forward transmissivity is 85%, the backward transmissivity is 0.1%. Electromagnetic field profiles reveal that the designed diode strongly supports surface plasmons coupled across counterintuitive metal-dielectric pixel arrays. Thereby, it yields the transmission of first-order diffracted light with a high amplitude. In contrast, backward transmission has decoupled surface plasmons that redirect Poynting vectors back to the incident medium, resulting in near attenuation of its transmission. In addition, we experimentally verify the optical isolation function of the optical diode.

The complete mitochondrial genome of the sand bubbler crab Scopimera longidactyla Shen 1932 (Decapoda: Dotillidae) and its phylogenetic analysis.

The sand bubbler crab, Scopimera longidactyla Shen, 1932 (Arthropoda: Malacostraca: Decapoda: Thoracotremata: Dotillidae), is commonly found along tropical and subtropical sandy shores of China, Korea, and Taiwan. Ecologically, it plays an important role in the productivity of sandy shores through their feeding and burrowing activities. In this study, the first complete mitochondrial genome (mitogenome) of S. longidactyla was analyzed using next-generation sequencer. Its mitogenome, circular in structure, spans 15,965 bp with a GC content of 29.97%, consisting of 13 protein-coding genes, two ribosomal RNA genes, 22 transfer RNA genes, and one putative control region. Its mitogenome arrangement and composition are identical to its two congeners, S. globosa and S. intermedia. Phylogenetic analysis fully supports for the monophyly of the genus Scopimera and the sister relationship between S. longidactyla and S. globosa. The complete mitogenome of S. longidactyla and its phylogenetic implications will provide valuable insights for further studies in phylogenetic and evolutionary biology.

Potential of Lycii Radicis Cortex as an Ameliorative Agent for Skeletal Muscle Atrophy.

Lycii Radicis Cortex (LRC) is a traditional medicine in East Asia with various beneficial effects, including antioxidant, anti-inflammatory, anti-tumor, anti-diabetic, and anti-depressant properties. However, its potential effects on skeletal muscle atrophy have not been studied. In this study, the protective effects of LRC extract (LRCE) on dexamethasone (DEX)-induced muscle atrophy were investigated in C2C12 myotubes and mice. We evaluated the effect of LRCE on improving muscle atrophy using a variety of methods, including immunofluorescence staining, quantitative polymerase chain reaction (qPCR), Western blot, measurements of oxidative stress, apoptosis, ATP levels, and muscle tissue analysis. The results showed that LRCE improved myotube diameter, fusion index, superoxide dismutase (SOD) activity, mitochondrial content, ATP levels, expression of myogenin and myosin heavy chain (MHC), and reduced reactive oxygen species (ROS) production in dexamethasone-induced C2C12 myotubes. LRCE also enhanced protein synthesis and reduced protein degradation in the myotubes. In mice treated with DEX, LRCE restored calf thickness, decreased mRNA levels of muscle-specific RING finger protein 1 (MuRF1) and atrogin-1, and increased insulin-like growth factor 1 (IGF-1) mRNA level. Moreover, LRCE also repaired gastrocnemius muscle atrophy caused by DEX. Although human studies are not available, various preclinical studies have identified potential protective effects of LRCE against muscle atrophy, suggesting that it could be utilized in the prevention and treatment of muscle atrophy.

Crosstalk between Signaling Pathways and Energy Metabolism in Pluripotency.

The sequential change from totipotency to multipotency occurs during early mammalian embryo development. However, due to the lack of cellular models to recapitulate the distinct potency of stem cells at each stage, their molecular and cellular characteristics remain ambiguous. The establishment of isogenic naïve and primed pluripotent stem cells to represent the pluripotency in the inner cell mass of the pre-implantation blastocyst and in the epiblast from the post-implantation embryo allows the understanding of the distinctive characteristics of two different states of pluripotent stem cells. This review discusses the prominent disparities between naïve and primed pluripotency, including signaling pathways, metabolism, and epigenetic status, ultimately facilitating a comprehensive understanding of their significance during early mammalian embryonic development.

Transformation of nanoparticles via the transition of functional DNAs responsive to pH and vascular endothelial growth factor for photothermal anti-tumor therapy.

This study presents a novel approach for the development of DNA-functionalized gold nanoparticles (AuNPs) capable of responding to disease-specific factors and microenvironmental changes, resulting in an effective anti-tumor effect via photothermal therapy. The AuNPs are decorated with two types of DNAs, an i-motif duplex and a VEGF split aptamer, enabling recognition of changes in pH and VEGF, respectively. The formation of VEGF aptamers on the AuNPs induces their aggregation, further enhanced by VEGF ligands. The resulting changes in the optical properties of the AuNPs are detected by monitoring the absorbance. Upon irradiation with a near-infrared laser, the aggregated AuNPs generate heat due to their thermoplasmonic characteristic, leading to an anti-tumor effect. This study demonstrates the enhanced anti-tumor effect of DNA-functionalized AuNPs via photothermal therapy in both in vitro and in vivo tumor models. These findings suggest the potential utilization of such functional AuNPs for precise disease diagnosis and treatment by detecting disease-related factors in the microenvironment.

Predicting synthesis recipes of inorganic crystal materials using elementwise template formulation.

While advances in computational techniques have accelerated virtual materials design, the actual synthesis of predicted candidate materials is still an expensive and slow process. While a few initial studies attempted to predict the synthesis routes for inorganic crystals, the existing models do not yield the priority of predictions and could produce thermodynamically unrealistic precursor chemicals. Here, we propose an element-wise graph neural network to predict inorganic synthesis recipes. The trained model outperforms the popularity-based statistical baseline model for the top-k exact match accuracy test, showing the validity of our approach for inorganic solid-state synthesis. We further validate our model by the publication-year-split test, where the model trained based on the materials data until the year 2016 is shown to successfully predict synthetic precursors for the materials synthesized after 2016. The high correlation between the probability score and prediction accuracy suggests that the probability score can be interpreted as a measure of confidence levels, which can offer the priority of the predictions.