Physical modeling and deep learning are known for their respective advantages in interpretability and computational efficiency. Nonetheless, efficiently predicting properties of metallic materials while maintaining interpretability presents a formidable challenge. This study proposes a novel solution by introducing dual-output deep learning model that simultaneously predicts stress-strain partitioning and mechanical properties through a two-component architecture. The initial component uses U-Net model trained on stress and strain partitioning generated from crystal plasticity (CP) simulations, thereby enhancing interpretability. Subsequently, this information is used to predict the properties in the second component. The prediction results demonstrate the validity of this approach, accurately predicting high stress at the martensite-martensite interface, high strain at the ferrite-martensite interface, and properties. In addition, the minimal computational cost significantly improves efficiency compared to conventional CP method. This innovative methodology represents a significant advancement, achieving harmonious balance between interpretability, computational accuracy, and efficiency in properties prediction of metallic materials.

The establishment of 'microstructure-property' relationship is a long-standing topic for metallic materials. For steels with complex phases and morphologies, it is often difficult to identify the key microstructural constituents for the accurate prediction of tensile properties. Here, instead of traditional methods that extract specific information from microstructure images (e.g. grain size and volume fractions), a deep learning (DL) strategy using microstructure images as direct inputs, is implemented to predict the tensile property of dual-phase (DP) steel. Compared with traditional physical models, the proposed method can help overcome the difficulty in quantifying complex microstructural information. Additionally, this method is based on a small sample database, which ensures its high portability and applicability. An important visualization heat map analysis is also used to quantitatively identify the key microstructural factors that influence tensile properties, and further improves the explicability of this method.

The establishment of composition-microstructure-property relationship is a long-standing topic in materials science, yet neither continuum mechanics approaches nor machine learning methods have ever established a generic model that can cover a wide spectrum of stress-strain relationships for metallic materials and, in particular, for steels having complex microstructures. In this study, a deep learning framework that handles a multimodal database, including composition and multi-source microstructure images, is developed for predicting the tensile properties of dual-phase steels. The model exhibits excellent generality for that it enables the coupling analysis of multimodal data. Moreover, the integration of multi-source microstructure information significantly enables the model to be compatible with multiple mechanisms and achieves accurate prediction in the large stress-strain range. Finally, the reverse visualization applied in the proposed framework deepens the mechanism understanding of strain distribution under different phase morphologies, which greatly improves the explicability of the deep learning model. This framework provides useful guidance for the prediction of composition-microstructure-property relationship in complex steel systems and can be potentially applied to other alloys.

Creep-oriented alloy design is a long-standing interesting topic in the field of metal structural materials. However, the high cost for creep testing limits the development efficiency of new alloys using traditional trial-and-error methods. Additionally, the complex mechanism and influencing factors significantly increase the difficulty of physical modeling and simulation-guided design. In this study, an alloy design framework for creep life improvement is established, including two modules: creep life prediction and high-throughput design. For the first module, based on the creep dataset used in this research, the best machine learning model for creep life prediction was obtained by comparison of various machine learning strategies. By this way, the limitation of complex creep mechanism was partially removed and an accurate and generic model was established. For the second module, a genetic algorithm with a filter was used to obtain promising new alloy plans with optimal composition and processing parameters under specific creep conditions. After proving the rationality of this design framework and related design results, map design was used to guide the new alloy development for different creep conditions, and a new multifunctional alloy plan was proposed. This design system could provide preliminary guidance for high-efficiency alloy designs with complex target properties.

The multi-attribute method (MAM) is a liquid chromatography-mass spectrometry (LC-MS)-based method that is used to directly characterize and monitor numerous product quality attributes (PQAs) at the amino acid level of a biopharmaceutical product. MAM enables identity testing based on primary sequence verification, detection and quantitation of post-translational modifications and impurities. This ability to simultaneously and directly determine PQAs of therapeutic proteins makes MAM a more informative, streamlined and productive workflow than conventional chromatographic and electrophoretic assays. MAM relies on proteolytic digestion of the sample followed by reversed-phase chromatographic separation and high-resolution LC-MS analysis in two phases. First, a discovery study to determine quality attributes for monitoring is followed by the creation of a targeted library based on high-resolution retention time plus accurate mass analysis. The second aspect of MAM is the monitoring phase based on the target peptide library and new peak detection using differential analysis of the data to determine the presence, absence or change of any species that might affect the activity or stability of the biotherapeutic. The sample preparation process takes between 90 and 120 min, whereas the time spent on instrumental and data analyses might vary from one to several days for different sample sizes, depending on the complexity of the molecule, the number of attributes to be monitored and the information to be detailed in the final report. MAM is developed to be used throughout the product life cycle, from process development through upstream and downstream processes to quality control release or under current good manufacturing practices regulations enforced by regulatory agencies.

During the manufacturing of therapeutic proteins, Critical Quality Attributes (CQAs) have been monitored by conventional methods, such as cation exchange chromatography (CEX), reduced capillary electrophoresis-sodium dodecyl sulfate (rCE-SDS), and 1,2-diamino-4,5-methylenedioxybenzene (DMB) labelling method. The conventional methods often generate individual peaks that contain multiple components, which may obscure the detection and the quantification of individual critical quality attributes (CQAs). Alternatively, Multi-Attribute Method (MAM) enables detection and quantification of specific CQAs. A high resolution MAM has been developed and qualified to replace several conventional methods in monitoring product quality attributes, such as oxidation, deamidation, clipping, and glycosylation. The qualified MAM was implemented in process characterization, as well as release and stability assays in quality control (QC). In combination with a design-of-experiments (DoE), the MAM method identified multivariate process parameter ranges that yield acceptable CQA level, which provides operational flexibility for manufacturing.

In the past decade, advances in both separations and mass spectrometry (MS) technologies have enabled new, streamlined, and data-rich approaches to monitor product quality attributes and their relationship with process parameters throughout the lifecycle of therapeutic proteins. As we enter a new decade of technology and method development, MS-based approaches utilized in the biopharmaceutical industry are evolving further. In this mini-review, we explore key developments that could inspire and improve the future of therapeutic protein development.

The multi-attribute method (MAM) was conceived as a single assay to potentially replace multiple single-attribute assays that have long been used in process development and quality control (QC) for protein therapeutics. MAM is rooted in traditional peptide mapping methods; it leverages mass spectrometry (MS) detection for confident identification and quantitation of many types of protein attributes that may be targeted for monitoring. While MAM has been widely explored across the industry, it has yet to gain a strong foothold within QC laboratories as a replacement method for established orthogonal platforms. Members of the MAM consortium recently undertook an interlaboratory study to evaluate the industry-wide status of MAM. Here we present the results of this study as they pertain to the targeted attribute analytics component of MAM, including investigation into the sources of variability between laboratories and comparison of MAM data to orthogonal methods. These results are made available with an eye toward aiding the community in further optimizing the method to enable its more frequent use in the QC environment.

Various models were established for deformation-induced martensite start temperature prediction over decades. However, most of them are empirical or considering limited factors. In this research, a dual mode database for medium Mn steels was established and a convolutional neural network model, which considered all composition, critical processing information and microstructure images as inputs, was built for Msσ prediction. By comprehensively considering composition, processing and microstructure factors, this model was more rational and much more accurate than traditional thermodynamic models. Also, by the full use of images information, this model has stronger ability to overcome overfitting compared with various traditional machine learning models. This framework provides inspiration for the similar data analysis issues with small sample datasets but different data modes in the field of materials science.

The Multi-Attribute Method (MAM) Consortium was initially formed as a venue to harmonize best practices, share experiences, and generate innovative methodologies to facilitate widespread integration of the MAM platform, which is an emerging ultra-high-performance liquid chromatography-mass spectrometry application. Successful implementation of MAM as a purityindicating assay requires new peak detection (NPD) of potential process- and/or product-related impurities. The NPD interlaboratory study described herein was carried out by the MAM Consortium to report on the industry-wide performance of NPD using predigested samples of the NISTmAb Reference Material 8671. Results from 28 participating laboratories show that the NPD parameters being utilized across the industry are representative of high-resolution MS performance capabilities. Certain elements of NPD, including common sources of variability in the number of new peaks detected, that are critical to the performance of the purity function of MAM were identified in this study and are reported here as a means to further refine the methodology and accelerate adoption into manufacturer-specific protein therapeutic product life cycles.

Rapid technological advances have significantly improved the capability, versatility, and robustness of mass spectrometers which has led to them playing a central role in the development, characterization, and regulatory filings of biopharmaceuticals. Their application spans the entire continuum of drug development, starting with discovery research through product development, characterization, and marketing authorization and continues well into product life cycle management. The scope of application extends beyond traditional protein characterization and includes elements like clone selection, cell culture physiology and bioprocess optimization, investigation support, and process analytical technology. More recently, advances in the MS-based multi-attribute method are enabling the introduction of MS in a cGMP environment for routine release and stability testing. While most applications of MS to date have been for monoclonal antibodies, the successes and learnings should translate to the characterization of next-gen biotherapeutics where modalities like multispecifics could be more prevalent. In this review, we describe the most significant advances in MS and correlate them to the broad spectrum of applications to biotherapeutic development. We anticipate rapid technological improvements to continue that will further accelerate widespread deployment of MS, thereby elevating our overall understanding of product quality and enabling attribute-focused product development.

The design of high-strength steel has long been discussed in the field of metal structural materials. To further increase the strength of common high-strength steel and further decrease the cost for production, three direct-quench hot rolled steels were designed and fabricated. The rolling and coiling processes were set based on continuous cooling transformation curves. In addition, the effect of the coiling temperature on the tensile properties was discussed to further guide the optimization of the process. It was found that compared with granular bainite, lower bainite probably has more advantages for both the strength and low temperature impact toughness of direct-quench hot rolled steels. Through a process of tailoring the morphology of bainite and controlling the grain boundary precipitation, the newly designed direct-quench hot rolled steels showed greatly improved strength and acceptable ductility/toughness compared with traditional quenched and tempered steels.

Mass spectrometry is a powerful analytical tool that has been widely used in the biotech industry. Recent trends of implementing a liquid chromatography-mass spectrometry (LC-MS) based multi-attribute method (MAM) in cGMP environments enables end-to-end monitoring and control of critical quality attributes (CQAs) of therapeutic proteins to ensure drug product quality. Mass spectrometry is a powerful analytical tool that has been widely used in the biotech industry. Recent trends of implementing a liquid chromatography-mass spectrometry (LC-MS) based multi-attribute method (MAM) in cGMP environments enables end-to-end monitoring and control of critical quality attributes (CQAs) of therapeutic proteins to ensure drug product quality.

In this commentary, we will provide a high-level introduction into LC-MS product characterization methodologies deployed throughout biopharmaceutical development. The ICH guidelines for early and late phase filings is broad so that it is applicable to diverse biotherapeutic products in the clinic and industry pipelines. This commentary is meant to address areas of protein primary sequence confirmation and sequence variant analysis where ambiguity exists in industry on the specific scope of work that is needed to fulfill the general guidance that is given in sections Q5b and Q6b. This commentary highlights the discussion and outcomes of two recent workshops centering on the application of LC-MS to primary structure confirmation and sequence variant analysis (SVA) that were held at the 2018 and 2019 CASSS Practical Applications of Mass Spectrometry in the Biotechnology Industry Symposia in San Francisco, CA and Chicago, IL, respectively. Recommendations from the conferences fall into two distinct but related areas; 1) consolidation of opinions amongst industry stakeholders on the specific definitions of peptide mapping and peptide sequencing for primary structure confirmation and the technologies used for both, as they relate to regulatory expectations and submissions and 2) development of fit-for-purpose strategy to define appropriate assay controls in SVA experiments.

Mass spectrometry is a powerful analytical tool that has been widely used in the biotech industry. Recent trends of implementing a liquid chromatography-mass spectrometry (LC-MS) based multi-attribute method (MAM) in cGMP environments enables end-to-end monitoring and control of critical quality attributes (CQAs) of therapeutic proteins to ensure drug product quality.

Proof of concept evidence is presented for a new method for the determination of isoaspartate, an important post-translational modification. Chemical derivatization is performed using common reagents for the modification of carboxylic acids and shown to yield suitable diagnostic information with regard to isomerization at the aspartate residue. The diagnostic gas phase chemistry is probed by collision-induced dissociation mass spectrometry, on the timescale of the MS experiment and semi-quantitative calibration of the percentage of isoaspartate in a peptide sample is demonstrated. Graphical Abstract ᅟ

Ethylene oxide (EtO) is widely used in sterilization of drug product primary containers and medical devices. The impact of residual EtO on protein therapeutics is of significant interest in the biopharmaceutical industry. The potential for EtO to modify individual amino acids in proteins has been previously reported. However, specific identification of EtO adducts in proteins and the effect of residual EtO on the stability of therapeutic proteins has not been reported to date. This paper describes studies of residual EtO with two therapeutic proteins, a PEGylated form of the recombinant human granulocyte colony-stimulating factor (Peg-GCSF) and recombinant human erythropoietin (EPO) formulated with human serum albumin (HSA). Peg-GCSF was filled in an EtO sterilized delivery device and incubated at accelerated stress conditions. Glu-C peptide mapping and LC-MS analyses revealed residual EtO reacted with Peg-GCSF and resulted in EtO modifications at two methionine residues (Met-127 and Met-138). In addition, tryptic peptide mapping and LC-MS analyses revealed residual EtO in plastic vials reacted with HSA in EPO formulation at Met-328 and Cys-34. This paper details the work conducted to understand the effects of residual EtO on the chemical stability of protein therapeutics.

A novel photo-oxidative cross-linking between two histidines (His-His) has been discovered and characterized in an IgG1 antibody via the workflow of XChem-Finder, (18)O labeling and mass spectrometry (Anal. Chem. 2013, 85, 5900-5908). Its structure was elucidated by peptide mapping with multiple proteases with various specificities (e.g., trypsin, Asp-N, and GluC combined with trypsin or Asp-N) and mass spectrometry with complementary fragmentation modes (e.g., collision-induced dissociation (CID) and electron-transfer dissociation (ETD)). Our data indicated that cross-linking occurred across two identical conserved histidine residues on two separate heavy chains in the hinge region, which is highly flexible and solvent accessible. On the basis of model studies with short peptides, it has been proposed that singlet oxygen reacts with the histidyl imidazole ring to form an endoperoxide and then converted to the 2-oxo-histidine (2-oxo-His) and His+32 intermediates, the latter is subject to a nucleophilic attack by the unmodified histidine; and finally, elimination of a water molecule leads to the final adduct with a net mass increase of 14 Da. Our findings are consistent with this mechanism. Successful discovery of cross-linked His-His again demonstrates the broad applicability and utility of our XChem-Finder approach in the discovery and elucidation of protein cross-linking, particularly without a priori knowledge of the chemical nature and site of cross-linking.

Insufficient copper level in the mammalian cell culture medium resulted in lactate accumulation while maintaining similar growth and culture viability profiles. Label-free, LC-MS/MS-based shotgun proteomics method was applied to compare the protein expression profiles obtained from the cultures exposed to suboptimal copper level to those provided with sufficient amount of copper. Under copper deficient condition, a substantial reduction of the protein levels of the multiple subunits of Complex IV, also known as cytochrome c oxidase, of the mitochondrial electron transport chain was observed for all three different Chinese Hamster Ovary (CHO) cell lines expressing therapeutic monoclonal antibodies tested. Additional proteins affected by suboptimal copper level included peroxiredoxin (PRDX) and hepatocyte-derived growth factor (HDGF), which were affected during early phase of the fed-batch production, several days prior to initiation of lactate accumulation. In contrast, proteins such as syntenin (SDCBP) and integral membrane 2C (ITM2C) showed altered expression patterns toward the end of culture duration, after lactate divergence had occurred. For all conditions tested, time was the most predominant factor facilitating the direction of global protein expression trend, with substantial number of proteins subjected to time-dependent changes in expression, independent of copper.