Chemical Information Research Articles

Advancements in hand-drawn chemical structure recognition through an enhanced DECIMER architecture

Accurate recognition of hand-drawn chemical structures is crucial for digitising hand-written chemical information in traditional laboratory notebooks or facilitating stylus-based structure entry on tablets or smartphones. However, the inherent variability in hand-drawn structures poses challenges for existing Optical Chemical Structure Recognition (OCSR) software. To address this, we present an enhanced Deep lEarning for Chemical ImagE Recognition (DECIMER) architecture that leverages a combination of Convolutional Neural Networks (CNNs) and Transformers to improve the recognition of hand-drawn chemical structures. The model incorporates an EfficientNetV2 CNN encoder that extracts features from hand-drawn images, followed by a Transformer decoder that converts the extracted features into Simplified Molecular Input Line Entry System (SMILES) strings. Our models were trained using synthetic hand-drawn images generated by RanDepict, a tool for depicting chemical structures with different style elements. A benchmark was performed using a real-world dataset of hand-drawn chemical structures to evaluate the model's performance. The results indicate that our improved DECIMER architecture exhibits a significantly enhanced recognition accuracy compared to other approaches.Scientific contributionThe new DECIMER model presented here refines our previous research efforts and is currently the only open-source model tailored specifically for the recognition of hand-drawn chemical structures. The enhanced model performs better in handling variations in handwriting styles, line thicknesses, and background noise, making it suitable for real-world applications. The DECIMER hand-drawn structure recognition model and its source code have been made available as an open-source package under a permissive license.Graphical

Microstructural analysis of ex-service neutron irradiated stainless steel nuclear fuel cladding by high-speed AFM

Fuel cladding in advanced gas-cooled nuclear reactors is made of an austenitic stainless steel referred to as 20Cr/25NiNb. This material can undergo microstructural changes in these extreme environments through a combination of thermal effects and radiation damage. In this work, the microstructures present in an ex-service irradiated 20Cr/25Ni-Nb sample are studied using correlated high-speed atomic force microscopy (HS-AFM) and electron microscopy techniques for the first time. Correlated topographic, crystallographic, and chemical information from the sample surface enabled identification of secondary phase precipitates (SPPs) including M23C6, sigma phase, NbC, and G phase. These SPPs can have adverse effects on the material properties. HS-AFM analysis showed surface textures unique to each SPP. Voids formed due to irradiation were also identified across the surface by HS-AFM. These voids were found to be larger in size and depth along grain boundaries. Further analysis identified a relationship between void size and frequency and grain boundary misorientation and SPP presence.

Quantitative Imaging of Magnesium Biodegradation by 3D X‐Ray Ptychography and Electron Microscopy

AbstractMagnesium‐based alloys are excellent materials for temporary medical implants, but understanding and controlling their corrosion performance is crucial. Most nanoscale corrosion studies focus on the surface, providing only 2D information. In contrast, macro‐ and microscale X‐ray tomography offers representative volume information, which is, however, comparatively low in resolution and rather qualitative. Here a new mesoscale approach overcomes these drawbacks and bridges the scale gap by combining 3D measurements using ptychographic X‐ray computed tomography (PXCT) with electron microscopy. This combination allows to observe the corrosion progress non‐destructively in 3D and provides high‐resolution chemical information on the corrosion products. A medical Mg–Zn–Ca alloy is used and compared the same sample in the pristine and corroded states. With PXCT an isotropic resolution of 85 and 123 nm is achieved for the pristine and corroded states respectively, which enables to distinguish nanoscale Mg2Ca precipitates from the matrix. The corroded state in best approximation to the in situ conditions is imaged and reveals the complexity of corrosion products. The results illustrate that the corrosion‐layer is dense and defect‐free, and the corrosion of the material is grain‐orientation sensitive. The developed workflow can advance research on bioactive materials and corrosion‐sensitive functional materials.

Chemistry in Externally FUV-irradiated Disks in the Outskirts of the Orion Nebula Cluster

Most stars are born in stellar clusters, and their protoplanetary disks, which are the birthplaces of planets, can, therefore, be affected by the radiation of nearby massive stars. However, little is known about the chemistry of externally irradiated disks, including whether or not their properties are similar to the so-far better-studied isolated disks. Motivated by this question, we present ALMA Band 6 observations of two irradiated Class II protoplanetary disks in the outskirts of the Orion Nebula Cluster to explore the chemical composition of disks exposed to (external) far-ultraviolet (FUV) radiation fields: the 216-0939 disk and the binary system 253-1536A/B, which are exposed to radiation fields of 102–103 times the average interstellar radiation field. We detect lines from CO isotopologues, HCN, H2CO, and C2H toward both protoplanetary disks. Based on the observed disk-integrated line fluxes and flux ratios, we do not find significant differences between isolated and irradiated disks. The observed differences seem to be more closely related to the different stellar masses than to the external radiation field. This suggests that these disks are far enough away from the massive Trapezium stars, that their chemistry is no longer affected by external FUV radiation. Additional observations toward lower-mass disks and disks closer to the massive Trapezium stars are required to elucidate the level of external radiation required to make an impact on the chemistry of planet formation in different kinds of disks.

MuMCyp_Net: A multimodal neural network for the prediction of Cyp450 inhibition

The cytochrome P450 (CYP450) enzymes, heme-containing monooxygenases, are indispensable in metabolizing a broad range of drugs and other xenobiotics. Nonetheless, specific chemical entities (molecules) can inhibit these enzymes, resulting in undesirable drug-drug interactions, alterations in pharmacokinetics, and potentially compromising the efficacy and safety of drugs. Therefore, it is critical to identify molecules exhibiting inhibitory activities toward the five majors human CYP450 enzymes (1A2, 2C9, 2C19, 2D6, and 3A4) in the preliminary phases of drug development. This study presents a novel deep-learning model, MuMCyp_Net, for predicting CYP450 inhibition by small molecules. The architecture of MuMCyp_Net integrates a convolutional neural network (CNN) with an Attention mechanism, a bi-directional gated recurrent unit (biGRU) network, and a deep neural network (DNN) to efficiently process local chemical context information and global molecular properties of molecules. A total of 25,753 distinct compound molecules were utilized to train and evaluate the model’s performance. The proposed model demonstrates competitive performance as indicated by matthew’s correlation coefficients (MCC), ranging from 0.63 to 0.68, an accuracy rate between 0.82 and 0.90, and an AUC of 0.86 to 0.92. This novel approach shows considerable promise in accurately identifying CYP450 inhibitors, contributing to safer and more effective drug development. To further utilize the potential of the proposed model, a freely accessible web server tool for the virtual screening of molecules was developed, which will aid in identifying CYP450 inhibitors from non-inhibitors. The web-based tool can be accessed at https://mumcypnet.streamlit.app/.

Application of design of experiment for quantification of 71 new psychoactive substances in influent wastewater

BackgroundNew psychoactive substances (NPS) are of public health concern due to their sporadic proliferation and the dearth of information on toxicity when consumed. In addition to seized data from forensic and toxicology reporting, wastewater analysis serves as a complimentary tool for NPS surveillance. A method to detect 71 NPS by simple filtration followed by liquid-chromatography tandem mass spectrometry was developed to detect multiclass NPS consisting of arylcyclohexylamines, designer benzodiazepines, synthetic cannabinoids, synthetic opioids, phenethylamines, synthetic cathinones, tryptamines, and indole alkaloids. ResultsIn this work, the influential factors for electrospray ionisation were identified and optimised using the fractional factorial design and face-centred central composite design, respectively. The filtration loss during sample clean-up was assessed for all compounds. The final method was validated and applied to wastewater collected from a music festival held in Queensland in 2022. The validated method had linearity between 0.5 ng L−1 and 5000 ng L−1, the limit of quantification (LOQ) ranges from 0.6 ng L−1 to 70 ng L−1, precision within ±20 %, accuracy ranges from 70 % to 120 %, and matrix effect ranges from soft (0 %–20 %) to medium (20 %–50 %) for the majority of the compounds. NPS detected in the festival were 2-fluorodeschloroketamine, 7-hydroxymitragynine, mitragynine, N,N-dimethylpentylone, pentylone, phenibut, and O-desmethyltramadol. SignificanceSystematic electrospray ionisation optimisation using the design of experiment for a large method is practical and provides in-depth chemical information on studied compounds. The optimised method demonstrated the applicability of analysing samples collected from a festival in this work.

Adaptively multi-scale microstructure characterization of cancellous bone via photoacoustic signal decomposition

Osteoporosis is a systemic disease with a high incidence in the elderly and seriously affects the quality of life of patients. Photoacoustic (PA) technology, which combines the advantages of light and ultrasound, can provide information about the physiological structure and chemical information of biological tissues in a non-invasive and non-radiative way. Due to the complex structural characteristics of bone tissue, PA signals generated by bone tissue are non-stationary and nonlinear. However, conventional PA signal processing methods are not effective for non-stationary signal processing. In this study, an empirical mode decomposition (EMD)-based Hilbert-Huang transform (HHT) PA signal analysis method, called HHT PA signal analysis (HPSA), was developed to assess the microstructure information of bone tissue, which is closely related to bone health. The feasibility of the HPSA method in bone health assessment was proven by numerical simulation and experimental studies on animal samples with different bone volume/total volume (BV/TV) and bone mineral densities. First, based on adaptive EMD, the different modes correlated with multi-scale information were mined from the PA signal, the correlations between different intrinsic mode function (IMF) modes and BV/TVs were analyzed, and the optimal mode for more efficient PA time–frequency analysis was selected. Second, multi-wavelength HPSA was used to assess the changes in the chemical components of the bone tissue. The results demonstrate that the HPSA method can distinguish bones with different BV/TVs and microstructure conditions adaptively with high efficiency. They further emphasize the potential of PA techniques in characterizing biological tissues in bones for early and rapid detection of bone diseases.

4D-STEM analysis of materials: Progress and Future

4D-STEM (four dimensional scanning transmission electron microscopy) is a powerful technique for analysis from which structural, physical, chemical information can be extracted. When it comes to 4D-STEM experiment, a converged electron beam is scanned across the 2D area of sample, and a pixelated detector records a 2D diffraction pattern at each scan position. Thus, the final datasets have four dimensions. This paper will cover material information that can be obtained through 4D STEM and summarize the current status and use of 4D-STEM studies including virtual imaging, phase retrieval (ptychography) and structural analysis of amorphous materials. The future perspective of 4D-STEM for new findings is also discussed.

How the secrets behind photocurrents are revealed in Ag-TiO2 heterostructures-based plasmonic photoelectrochemical systems: A collaborative approach of EC-SERS and photoelectrochemical methods

Plasmon-mediated chemical reactions (PMCR) have garnered growing interest as a promising concept for photocatalysis. However, in electrochemical systems at solid–liquid interfaces, the photo-induced charge transfer on the surface of metal–semiconductor heterostructures involves complex processes and mechanisms, which are still poorly understood. We explore the plasmon-mediated carrier transfer mechanism and the synergistic effect of light and electric fields on Ag-TiO2 heterostructures, through a combination of electrochemical surface-enhanced Raman spectroscopy and photoelectrochemical methods, with para-aminothiophenol (PATP) serving as a probe molecule. The results show that photocurrent responses are dependent on not only excitation wavelengths and applied potentials, but also the irreversibility of redox. The relationship between photocurrent responses and the chemical transformation between PATP and 4,4′-dimercaptoazobenzene is established, reflecting the photo-induced charge transfer of the heterostructures. The collaboration of spectroscopic and photoelectrochemical methods provide valuable insights into the chemical transformation and kinetic information of adsorbed molecules on the heterostructure during PMCR, offering opportunities for modulating of photocatalytic activities of hot carriers.

Biochar data into structure: A methodology for generating large-scale atomistic representations

A well-defined methodology for constructing appropriate atomistic representations of biochar will aid in visualizing the structural features and elucidating biochar behavior with molecular dynamics (MD) simulations. Such knowledge will facilitate engineering biochars tailored to specific applications. To achieve this goal, we adapted modeling strategies applied in coal science by employing multi-cross-polarization 13C nuclear magnetic resonance, ultimate analysis, Fourier-transform infrared spectroscopy, and X-ray photoelectron spectroscopy to identify functional groups. Helium density, surface area, and porosity were used to assess structural features. Biochar's aromatic cluster size distribution was proposed based on data from the benzene polycarboxylic acid method. The computational framework reduces bias by incorporating chemical information derived from density functional theory, reactive MD simulations, and advanced characterization data. The construction approach was successfully applied to cellulose biochars produced at four temperatures, obtaining independent representations with a relative error on the atomic contents of <10 % for oxygen and nitrogen and <5 % for carbon and hydrogen. The atomistic representations were validated using X-ray diffraction, electron spin resonance data, and laser desorption/ionization Fourier-transform ion cyclotron resonance-mass spectrometry. The code will assist others in overcoming structural creation barriers and enable the utilization of the generated structures for further simulations.

Insights into Heterogeneous Catalysis by Tip‐Enhanced Raman Spectroscopy

AbstractHeterogeneous catalysis is immensely important in the chemical industry. Understanding the molecular mechanism of heterogeneous catalytic processes is fundamental to the design of high‐performance catalysts and the development of energy conversion technology. Tip‐enhanced Raman spectroscopy (TERS) provides invaluable topographic and chemical information with nanometer‐scale spatial resolution, thus making it a promising tool for the in‐depth characterization of heterogeneous catalytic processes. In this perspective, we first introduce the background of heterogeneous catalysis and TERS, and then we systematically discuss the application of TERS in heterogeneous catalysis. Key insights related to catalyst configurations, catalytically active sites, reacting species, spillover of active species, and structure‐reactivity relationship were summarized. Finally, we discuss the challenges and future development in the field.

Multi-excitation photoluminescence spectroscopy system for gemstone analysis.

Luminescence spectra can reveal important chemical and structural information that can be used for gemstone characterization and identification. Traditionally, gemstone UV-excited luminescence is evaluated visually under mercury vapor lamp illumination. This approach is limited by several factors, including the mixture of mercury's emission peaks, possible filter degradation, an inability to separate overlapping emission features, and the sensitivity and subjectivity of human vision and color interpretation. A multi-excitation photoluminescence (PL) spectroscopy system has been built for gemstone analysis, incorporating 261 and 405nm laser excitations to study gemstone emission features between 270 to 1000nm. This system presents significant improvements, extending the detection spectral range, increasing the sensitivity, accuracy and reproducibility of gemstone luminescence analysis. Luminescence analysis of commercially valuable gemstones are presented to demonstrate the system's suitability for gemstone identification. Examples include distinguishing natural from laboratory-grown diamonds, thermal and color treatment detection for corundum and pearls, respectively, and mineral type separation of emeralds and other green gemstones.

Exploring uncommon Fe-oxides in non-metallic inclusions in ultra-high-strength steel

Investigating non-metallic inclusions within ultra-high-strength-steel via conventional methods is a known: however, the challenge is to obtain chemical information of such inclusions at the sub-micrometer level. In this context, probing Fe-based oxide in inclusions is a vital aspect for guiding steel’ performance. The vibrational properties of sub micrometer size Fe-based oxides were investigated by Raman mapping along with chemometric analysis with the aim of probing their chemical composition. Highly contrasted Raman spectra were recorded from several inclusions embedded at different spatial locations. The observed spectral features were identified as specific markers of hematite (α-Fe2O3) and magnetite (Fe3O4). Principal Component Analysis was used to confirm the presence of these markers and potentially revealing additional patterns. Their unambiguous assignment has been inferred by comparing our experimental findings with the literature data recorded either in single crystals of iron oxides or oxyhydroxides. Micro-Raman spectroscopy is proven to be a reliable, cost-effective, and non-invasive tool for the unambiguous identification of subsurface regions of steel.

Radiolytic reduction of silver nanoparticles probed in situ via Cherenkov spectroscopy

Expanding on our previous efforts demonstrating the feasibility of extracting chemical information in steady-state systems of nanoparticles under irradiation, we demonstrate that the Cherenkov spectra can be used to observe chemical changes in real-time, in situ, and for non-equilibrium systems under irradiation. This is done by way of a dose-resolved measurement of the localized plasmon resonance absorption band associated with the formation and growth of silver nanoclusters and nanoparticles produced via radiolytic reduction of a solution of silver ion under exposure to an electron beam. Additionally, the method utilized in this study provided a significant resolution with a simplified optical system. This paves the way for a technique to determine similar changes in the material in medicinal, industrial (nuclear industry), and research applications—and indeed in any application involving ionizing radiation.

Technical note: Determining chemical composition of atmospheric single particles by a standard-free mass calibration algorithm

Abstract. The chemical composition of individual particles can be revealed by single-particle mass spectrometers (SPMSs). With higher accuracy in the ratio of mass to charge (m/z), more detailed chemical information could be obtained. In SPMSs, the conventional standard-based calibration methods (internal/external) are constrained by the inhomogeneity of ionization lasers and the finite focusing ability of the inlet system, etc.; therefore, the mass accuracy is restricted. In this study, we obtained the detailed and reliable chemical composition of single particles utilizing a standard-free mass calibration algorithm. In the algorithm, the characteristic distributions of hundreds of ions were concluded and collected in a database denoted as prototype. Each single-particle mass spectrum was initially calibrated by a function with specific coefficients. The range of coefficients was constrained by the magnitude of mass deviation to a finite vector space. To find the optimal coefficient vector, the conformity of each initially calibrated spectrum to the prototype dataset was assessed. The optimum calibrated spectrum was obtained with maximum conformity. For more than 98 % ambient particles, a 20-fold improvement in mass accuracy, from ∼ 10 000 ppm (integer) to ∼ 500 ppm (two decimal places), was achieved. The improved mass accuracy validated the determination of adjacent ions with a m/z difference ∼ 0.05 Th. Furthermore, atmospheric particulate trace elements that were poorly studied before are specified. The obtained detailed single-particle-level chemical information could help explain the source apportionment, reaction mechanism, and mixing state of atmospheric particles.

Exploring Enzyme-Mimicking Metal-Organic Frameworks for CO2 Conversion through Vibrational Spectra-Based Machine Learning.

In pursuing the benefits of natural enzyme catalysts while overcoming their limitations, we find metal-organic frameworks (MOFs), renowned for their highly tunable functionalities, stand out in biomimetic applications. We used unsupervised machine learning on density functional theory-computed vibrational infrared and Raman spectral features to screen 300 Zn-MOFs for CO2 conversion, similar to carbonic anhydrase (CA). Our findings confirmed that MOFs with spectroscopic attributes closely resembling those of CA hold the potential for replicating CA's electronic and catalytic properties. Unlike previous studies that relied on heuristic or trial-and-error methods and focused on geometric configurations, our research uses vibrational spectral features to explore structure-property relationships, making them more accessible through spectroscopy. Moreover, we highlight vibrational spectral features as efficient carriers for highly dimensional chemical information, enabling the simultaneous optimization of multiple performance parameters. These findings pave the way for pioneering designs of enzyme-mimetic MOFs and concurrently expand the application scope of spectroscopic tools in biomimetic catalysis.

Rapid identification of living cancer cells based on label-free surface-enhanced Raman spectroscopy

Surface-enhanced Raman spectroscopy (SERS) is a highly sensitive vibration spectroscopy technique that can obtain fingerprint information of biomolecules in cells, and it has demonstrated significant potential in the early diagnosis of cancer. Here, we propose a rapid and efficient identification method of living cancer cells (within 10 min), which is based on the label-free SERS detection using iodide-modified Ag nanoparticles (Ag IMNPs) and the ultra-fast Raman mapping by line illumination. The Ag IMNPs achieve a better balance between sensitivity, reproducibility, and biocompatibility when compared to conventional Au or Ag nanoparticles. Statistical analysis of Raman peak frequency and intensity across substantial SERS spectra reveals a wealth of intracellular chemical information about the three cell lines 7701/7703/Caski. This method can not only accurately identify and classify normal cells, cancer cells, and various types of cancer cells, but it can also detect fingerprint changes in the distribution of intracellular chemicals. Furthermore, we discover that, among the three cell lines, liver cancer cells 7703 had the highest level of reactive oxygen species (ROS). This rapid SERS method can provide a new optical detection method for early screening, progression monitoring and prognosis assessment of tumors.

Accurate prediction of drug combination risk levels based on relational graph convolutional network and multi-head attention

BackgroundAccurately identifying the risk level of drug combinations is of great significance in investigating the mechanisms of combination medication and adverse reactions. Most existing methods can only predict whether there is an interaction between two drugs, but cannot directly determine their accurate risk level.MethodsIn this study, we propose a multi-class drug combination risk prediction model named AERGCN-DDI, utilizing a relational graph convolutional network with a multi-head attention mechanism. Drug-drug interaction events with varying risk levels are modeled as a heterogeneous information graph. Attribute features of drug nodes and links are learned based on compound chemical structure information. Finally, the AERGCN-DDI model is proposed to predict drug combination risk level based on heterogenous graph neural network and multi-head attention modules.ResultsTo evaluate the effectiveness of the proposed method, five-fold cross-validation and ablation study were conducted. Furthermore, we compared its predictive performance with baseline models and other state-of-the-art methods on two benchmark datasets. Empirical studies demonstrated the superior performances of AERGCN-DDI.ConclusionsAERGCN-DDI emerges as a valuable tool for predicting the risk levels of drug combinations, thereby aiding in clinical medication decision-making, mitigating severe drug side effects, and enhancing patient clinical prognosis.

PubChem synonym filtering process using crowdsourcing

PubChem (https://pubchem.ncbi.nlm.nih.gov) is a public chemical information resource containing more than 100 million unique chemical structures. One of the most requested tasks in PubChem and other chemical databases is to search chemicals by name (also commonly called a “chemical synonym”). PubChem performs this task by looking up chemical synonym-structure associations provided by individual depositors to PubChem. In addition, these synonyms are used for many purposes, including creating links between chemicals and PubMed articles (using Medical Subject Headings (MeSH) terms). However, these depositor-provided name-structure associations are subject to substantial discrepancies within and between depositors, making it difficult to unambiguously map a chemical name to a specific chemical structure. The present paper describes PubChem’s crowdsourcing-based synonym filtering strategy, which resolves inter- and intra-depositor discrepancies in synonym-structure associations as well as in the chemical-MeSH associations. The PubChem synonym filtering process was developed based on the analysis of four crowd-voting strategies, which differ in the consistency threshold value employed (60% vs 70%) and how to resolve intra-depositor discrepancies (a single vote vs. multiple votes per depositor) prior to inter-depositor crowd-voting. The agreement of voting was determined at six levels of chemical equivalency, which considers varying isotopic composition, stereochemistry, and connectivity of chemical structures and their primary components. While all four strategies showed comparable results, Strategy I (one vote per depositor with a 60% consistency threshold) resulted in the most synonyms assigned to a single chemical structure as well as the most synonym-structure associations disambiguated at the six chemical equivalency contexts. Based on the results of this study, Strategy I was implemented in PubChem’s filtering process that cleans up synonym-structure associations as well as chemical-MeSH associations. This consistency-based filtering process is designed to look for a consensus in name-structure associations but cannot attest to their correctness. As a result, it can fail to recognize correct name-structure associations (or incorrect ones), for example, when a synonym is provided by only one depositor or when many contributors are incorrect. However, this filtering process is an important starting point for quality control in name-structure associations in large chemical databases like PubChem.

The phase changes of the mortars containing waste glass powder during carbonation

The utilization of glass powder (GP) with high pozzolanic activity to replace cement is a promising option for reusing waste glass. In such systems, Carbonation is a key factor that affects the durability. However, carbonation mechanisms remain poorly understood. In this study, the phase evolution of the GP - cement composites exposed to CO2 was determined by accelerated carbonation tests of samples with different cement-to-GP substitution rates (0, 15, 30, and 50 wt%). The carbonation depth was determined using phenolphthalein spray tests. Furthermore, the chemical information regarding the carbonation profiles was obtained by the X-ray diffraction (XRD), thermogravimetric analysis (TG), and Fourier transform infrared spectroscopy (FT-IR). It was found that an increase in the GP content increased the carbonation depth of the samples because of the low calcium hydroxide (CH) content. The increase rate was more similar to those of fly ash (FA) and silica fume (SF) rather than that of Ca-rich granulated blast-furnace slag (GBS). Moreover, XRD and TG tests revealed that compared to the plain cement, GP addition influenced crystalline forms of CaCO3, as well as the quantity of produced CaCO3. Combined with the FT-IR test results, when the replacement level of the GP up to ≥30 wt%, the produced CaCO3 and the depolymerised C-S-H by the carbonation significantly differed from the pure Portland cement. This study offers insights into the carbonation mechanisms of GP – cement mortars to promote glass recycling.