Assessing the feasibility of statistical inference using synthetic antibody-antigen datasets.

High-Resolution Mapping of B-Cell Epitopes on the Factor VIII C2 Domain Using Surface Plasmon Resonance

Lectins are proteins that bind glycoproteins; binding patterns are altered with changes in glycoprotein expression accompanying maturation or disease. Binding of two lectins, wheat germ agglutinin (WGA) and peanut agglutinin (PNA), in human and rodent colon were previously examined. Normal tissue showed intense WGA binding; PNA binding was minimal. Diseased tissues showed increased PNA binding. We hypothesized that N-(2-hydroxypropyl)methacrylamide (HPMA) copolymer-lectin-drug conjugates could deliver therapeutic agents to diseased tissues by targeting colonic glycoproteins. We examined biorecognition of free and HPMA copolymer-conjugated WGA and PNA and anti-Thomsen-Friedenreich (TF) antigen antibody binding in normal neonatal, adult, and diseased rodent tissues, human specimens of inflammation, and Barrett's esophagus. Neonatal WGA binding was comparable to the adult, with additional luminal columnar cell binding. PNA binding was more prevalent; luminal columnar cell binding existed during the first 2.5 weeks of life. WGA binding was strong in both normal and diseased adult tissues; a slight decrease was noted in disease. PNA binding was minimal in normal tissues; increases were seen in disease. Anti-TF antigen antibody studies showed that PNA did not bind to the antigen. The results suggest that HPMA copolymer-lectin-drug conjugates may provide site-specific treatment of conditions such as colitis and Barrett's esophagus.

Antibodies are versatile therapeutic molecules that use combinatorial sequence diversity to cover a vast fitness landscape. Designing optimal antibody sequences, however, remains a major challenge. Recent advances in deep learning provide opportunities to address this challenge by learning sequence–function relationships to accurately predict fitness landscapes. These models enable efficient in silico prescreening and optimization of antibody candidates. By focusing experimental efforts on the most promising candidates guided by deep learning predictions, antibodies with optimal properties can be designed more quickly and effectively. Here we present AlphaBind, a domain-specific model that uses protein language model embeddings and pre-training on millions of quantitative laboratory measurements of antibody–antigen binding strength to achieve state-of-the-art performance for guided affinity optimization of parental antibodies. We demonstrate that an AlphaBind-powered antibody optimization pipeline can deliver candidates with substantially improved binding affinity across four parental antibodies (some of which were already affinity-matured) and using two different types of training data. The resulting candidates, which include up to 11 mutations from parental sequence, yield a sequence diversity that allows optimization of other biophysical characteristics, all while using only a single round of data generation for each parental antibody. AlphaBind weights and code are publicly available at: https://github.com/A-Alpha-Bio/alphabind.

Habitat suitability index (HSI) models for seven fish guilds in two segments of the upper Roanoke River drainage, Virginia, were formulated for the summer seasons of 1989 and 1990. We considered five habitat variables as potential limiting factors: depth, average and demersal velocities, average substratum size, and percent cover. These physical variables were modeled both separately and as composite HSI indices. Composite models were built from linear regression equations (both simple and multiple) in which the observed guild density in quadrats was regressed against physical microhabitat variables or individual suitability indices (SIs = predicted fish densities). There were five major findings. First, habitat variables were used independently by most fish guilds, as statistical interactions were weak and inconsistent for regression models predicting guild densities from physical variables. That is, fish-microhabitat relations for target habitat variables were typically unaffected by the condition (value) of other habitat variables. Although polynomial (curvilinear) terms were stronger than interaction terms, linear terms accounted for most of the variation in guild densities among quadrats. Second, the predictive power of these complex physical models for guild densities was matched by that of multiplying the SIs for individual microhabitat variables together. Third, this product (joint-suitability-factor) approach was superior to other methods of developing composite HSIs from individual SIs because it was consistently accurate across fish guilds (owing to the lack of strong statistical interactions) and was a simpler regression model (involving only one slope coefficient). Fourth, observed guild densities for each river segment were well correlated with those predicted by the product equation with SI data from the other river segment, thus cross-validating our HSI models in the upper Roanoke River drainage. Fifth, maximum guild densities for habitat variables that were stratified into a few or several categories provided useful indices of the limiting factors for fish guilds because higher densities indicated greater habitat specialization. Across all guilds, depth was consistently the most important factor in habitat selection. In sum, our results suggest that fish-habitat statistical interactions are not strong enough to invalidate the product equation traditionally used by fish researchers to build composite HSI models, at least when SI data are aggregated by habitat use guild.

Larger, more diverse antibody libraries are being screened in recent years. Despite enabling the high-throughput screening of extensive libraries, the process for identification and characterization of antibodies remains laborious and time-intensive. Here we report a platform that not only compiles hit finding and validation processes, but also provides high-resolution binding affinity as well as rank order among the library being screened. We streamlined the library construction and affinity screening via microscale transient expression in HEK293T cells with designed antibody sequence and direct quantification of antibody-antigen binding affinity which reduced the timeline from multiple weeks to a single week. We constructed the single-chain fragment antigen binding (scFab) libraries by replacing the duplex DNA fragment in CDR of interest while preserving the human framework of Fab. We have demonstrated that the SPID technique allows to detect antibody-antigen interactions from just dozens of pg of scFabs in total. This allowed us to bypass the conventional DNA cloning or even PCR amplification to achieve the minimum amount of DNA plasmids for expressing scFabs to be analyzed such as BLI and SPR. Therefore, we assembled the DNA plasmids through in vitro ligation with high efficiency and directly introduced it to HEK293T in 96-well microplate with microscale (200 ul) for transient transfection, which only takes three days for library generation. Next, we utilized SPID to quantify the number of single molecules between free and antigen-bound fractions of scFab, resulting in occupancy at the antibody-antigen binding equilibrium. This technique facilitated the determination of binding affinity (KD) for 200 antibody-antigen pairs within a 6-hour TAT encompassing both measurement and analysis. In total, we utilized 4,000 times less antibody for characterization while achieving enhanced sensitivity compared to both BLI and SPR methods. As a proof-of-concept, we generated a library consisting of 720 variants of Trastuzumab in scFab format to probe the landscape of CDRs affecting affinities against HER2 showing nM to low uM dynamic range of KD. Out of 720 analyzed variants, 5% were deleterious, 38% provided substantial changes, 40% showed little/no change, 17% enhanced affinity to HER2. To validate the robustness of the platform, we extensively expanded the library screening, assessing over 2,000 antibody variants for their affinity against more than three different therapeutic targets. To conclude, we demonstrated an efficient approach for monitoring changes in affinity resulting from single mutations in scFab. This rapid screening of site-saturated libraries would lead to a combinational variant library that our platform can screen. Ultimately, our platform aims to identify potential candidates for therapeutic antibody development throughput the phases of antibody discovery, maturation, and engineering. Citation Format: Jihye Jo, Changju Chun, Byeong-Kwon Sohn, Booyoung Yu, Jiyu Lee. SMASH: Single molecule antibody screening with high-throughput imaging system [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3107.

An integral component of human spermatozoa, a glycoprotein of Mr 143,000 (two subunits of Mr 76,000 and 67,000) was recognized by the a-HS 1A.1 monoclonal antibody. The antigen was localized on the plasma membrane over the sperm head, as demonstrated by transmission electron microscopy. The antigen-antibody binding on gametes during changes in their functional state was followed by an indirect immunofluorescence assay of live human spermatozoa. In freshly ejaculated spermatozoa the antibody binding pattern revealed a patchwork quilt-like topography of the plasma membrane over the acrosome; the percentage of positive cells varied from 20 to 78% with a mean of 50% (n = 82). Incubation in a capacitation medium could increase this percentage up to 98%, revealing new epitopes in an energy-dependent and temperature-independent manner; concomitantly, a part of the antigen migrated in energy-independent and temperature-dependent manner and accumulated in a ring over the postacrosome. When an acrosome reaction was induced in vitro in the presence of Ca2+ with either A23187, ionomycin or human follicular fluid, the HS 1A.1 antigen migrated until immobilization in a well defined pattern around the equatorial segment (single band) or around the equatorial and postacrosomal segments (2 or, seldom, 3 bands). The new antigen localization resulted from a lateral diffusion of pre-existing molecules, occurred in only a few minutes, did not require energy and was temperature-dependent. At the same time, the well outlined large patch burst into a multitude of small spots before vanishing. this veil-like labelling was often observed in spermatozoa kept in the seminal plasma or treated with a metabolic poison. The HS 1A.1 antigen localization reflects surface changes induced by the incubation in a capacitation medium and the acrosome reaction. Apart from the regional heterogeneity of the plasma membrane of a single cell, as noted above, there were differences in the plasma membrane changes in individual spermatozoa from the same ejaculate as well as in semen samples from different donors. The new antibody binding pattern was often alike in successive ejaculates of the same donor. In patients consulting for infertility the percentage of positive cells was often low and migration of the antigen was slight or absent.

Free-Energy Simulations Reveal that Both Hydrophobic and Polar Interactions Are Important for Influenza Hemagglutinin Antibody Binding

Germ-free (GF) animals lack a colonic microflora like that seen in conventional (CV) animals. Bacterial presence plays a role in the development of glycoproteins in the gastrointestinal (GI) tract; the absence of a microbiota has been seen to suppress the production of certain glycoproteins and glycolipids. Binding patterns of lectins are modified when glycoprotein structures are altered (e.g., during development or disease). Little information on lectin binding patterns in mature GF animals is available. We examined the binding of free and N-(2-hydroxypropyl)methacrylamide (HPMA) copolymer-conjugated fluorescein isothiocyanate (FITC)-labeled wheat germ agglutinin (WGA) [P(HPMA)-(WGA-FITC)] and FITC-labeled peanut agglutinin (PNA) [P(HPMA)-(PNA-FITC] in CV and GF mouse colon with and without neuraminidase pretreatment. Anti-Thomsen-Friedenreich (TF) antigen (a development and disease-related glycoprotein) antibody binding was also examined in these tissues. Subtle differences were seen in the binding patterns between CV and GF animals. CV animals showed strong P(HPMA)-(WGA-FITC) binding in goblet cells, but minimal P(HPMA)-(PNA-FITC) binding was visible. In GF animals, luminal surface binding of P(HPMA)-(WGA-FITC) was visible, and goblet cell binding of P(HPMA)-(PNA-FITC) was seen. These subtle changes suggest that altered glycoprotein expression occurred under GF conditions.

Objectives. The aim of the study was to examine factors which determine stress coping styles in type 2 diabetic (T2D) patients, with regard to selected demographic variables, clinical diabetes-related variables and selected psychical variables (anxiety level and assessment of depressive disorders). Methods. 50 T2D patients, aged 59.9 ± 10.2 years were assessed by Coping Inventory for Stressful Situations (CISS), Spielberger State-Trait Anxiety Inventory (STAI), and Beck Depression Inventory (BDI). In the statistical analysis simple and multivariable logistic regression models were used. Results. Variables significantly increasing the selection risk of stress coping style different from preferred task-oriented strategy in a simple logistic regression model are: hypoglycemia within three months prior to the research: odds ratio (OR) = 6.86 (95% confidence interval (CI) 1.25–37.61), taking antidepressants or neuroleptics: OR =15.42 (95% CI 2.42–98.33), severe depression in Beck's scale: OR = 84.00 (95% CI 6.51–1083.65), high state-anxiety level: OR = 9.60 (95% CI 1.08–85.16), and high trait-anxiety level: OR = 18.40 (95%CI 2.96–114.31), but in a multivariable model, diagnosed depression is the strongest factor: OR = 32.38 (95% CI 4.94–212.13). Conclusions. In T2D patients, the strategy to cope with stress appears to be mostly influenced by psychical predisposition.

In medical imaging, challenges are competitions that aim to provide a fair comparison of different methodologic solutions to a common problem. Challenges typically focus on addressing real-world problems, such as segmentation, detection, and prediction tasks, using various types of medical images and associated data. Here, we describe the organization and results of such a challenge to compare machine-learning models for predicting survival in patients with diffuse large B-cell lymphoma using a baseline 18F-FDG PET/CT radiomics dataset. Methods: This challenge aimed to predict progression-free survival (PFS) in patients with diffuse large B-cell lymphoma, either as a binary outcome (shorter than 2 y versus longer than 2 y) or as a continuous outcome (survival in months). All participants were provided with a radiomic training dataset, including the ground truth survival for designing a predictive model and a radiomic test dataset without ground truth. Figures of merit (FOMs) used to assess model performance were the root-mean-square error for continuous outcomes and the C-index for 1-, 2-, and 3-y PFS binary outcomes. The challenge was endorsed and initiated by the Society of Nuclear Medicine and Molecular Imaging AI Task Force. Results: Nineteen models for predicting PFS as a continuous outcome from 15 teams were received. Among those models, external validation identified 6 models showing similar performance to that of a simple general linear reference model using SUV and total metabolic tumor volumes (TMTV) only. Twelve models for predicting binary outcomes were submitted by 9 teams. External validation showed that 1 model had higher, but nonsignificant, C-index values compared with values obtained by a simple logistic regression model using SUV and TMTV. Conclusion: Some of the radiomic-based machine-learning models developed by participants showed better FOMs than did simple linear or logistic regression models based on SUV and TMTV only, although the differences in observed FOMs were nonsignificant. This suggests that, for the challenge dataset, there was limited or no value seen from the addition of sophisticated radiomic features and use of machine learning when developing models for outcome prediction.

BackgroundTreatment with effective antiretroviral therapy (ART) lowers morbidity and mortality among HIV positive individuals. Effective highly active antiretroviral therapy (HAART) should lead to undetectable viral load within 6 months of initiation of therapy. Failure to achieve and maintain viral suppression may lead to development of resistance and increase the risk of viral transmission. In this paper three logistic regression based machine learning approaches are developed to predict early virological outcomes using easily measurable baseline demographic and clinical variables (age, body weight, sex, TB disease status, ART regimen, viral load, CD4 count). The predictive performance and generalizability of the approaches are compared.MethodsThe multitask temporal logistic regression (MTLR), patient specific survival prediction (PSSP) and simple logistic regression (SLR) models were developed and validated using the IDI research cohort data and predictive performance tested on an external dataset from the EFV cohort. The model calibration and discrimination plots, discriminatory measures (AUROC, F1) and overall predictive performance (brier score) were assessed.ResultsThe MTLR model outperformed the PSSP and SLR models in terms of goodness of fit (RMSE = 0.053, 0.1, and 0.14 respectively), discrimination (AUROC = 0.92, 0.75 and 0.53 respectively) and general predictive performance (Brier score= 0.08, 0.19, 0.11 respectively). The predictive importance of variables varied with time after initiation of ART. The final MTLR model accurately (accuracy = 92.9%) predicted outcomes in the external (EFV cohort) dataset with satisfactory discrimination (0.878) and a low (6.9%) false positive rate.ConclusionMultitask Logistic regression based models are capable of accurately predicting early virological suppression using readily available baseline demographic and clinical variables and could be used to derive a risk score for use in resource limited settings.

Cell phones have increased as a new communication technology in the modern world. To determine whether smartphone addiction is significantly associated with depression, anxiety, and stress among university nursing students in Acapulco, Guerrero, Mexico. This descriptive and cross-sectional study involved 212 students who voluntarily participated. Data were collected using two questionnaires: the Smartphone Addiction Scale Short Version (SAS-SV) and the Depression Anxiety and Stress Scale (DASS-21). The information was then analyzed using descriptive statistics and linear and simple logistic regression models. 46.70% (99) use their phones for more than 5 hours a day, and 38.20% (68) of the students presented smartphone addiction. Simple linear regression models showed a significant association between SAS-SV scores and DASS- 21 subscale scores. Simple logistic regression models indicated that students with cell phone addiction are 2.57 times more likely to suffer from depression, 2.50 times more likely to experience anxiety, and 3.34 times more likely to suffer from stress compared to those without cell phone addiction. Cell phone addiction was associated with such mental disorders among Mexican university students. These results could assist educational authorities in developing and implementing strategies to prevent depression, anxiety, and stress associated with smartphone use.

Generative machine learning (ML) has been postulated to become a major driver in the computational design of antigen-specific monoclonal antibodies (mAb). However, efforts to confirm this hypothesis have been hindered by the infeasibility of testing arbitrarily large numbers of antibody sequences for their most critical design parameters: paratope, epitope, affinity, and developability. To address this challenge, we leveraged a lattice-based antibody-antigen binding simulation framework, which incorporates a wide range of physiological antibody-binding parameters. The simulation framework enables the computation of synthetic antibody-antigen 3D-structures, and it functions as an oracle for unrestricted prospective evaluation and benchmarking of antibody design parameters of ML-generated antibody sequences. We found that a deep generative model, trained exclusively on antibody sequence (one dimensional: 1D) data can be used to design conformational (three dimensional: 3D) epitope-specific antibodies, matching, or exceeding the training dataset in affinity and developability parameter value variety. Furthermore, we established a lower threshold of sequence diversity necessary for high-accuracy generative antibody ML and demonstrated that this lower threshold also holds on experimental real-world data. Finally, we show that transfer learning enables the generation of high-affinity antibody sequences from low-N training data. Our work establishes a priori feasibility and the theoretical foundation of high-throughput ML-based mAb design.

Background and ObjectivesZZFrontal recess anatomy can be very complex, with accessory cells extending to the frontal sinus and possibly contributing to the obstruction of the frontal sinus. However, there is still controversy on the effect of the frontal recess cells. We designed this study to assess the effect of frontal recess cells on frontal sinusitis. Subjects and MethodZZWe retrospectively reviewed chart and collected data of those who visited the outpatient clinic between January and June, 2011. Parnasal sinus CT was taken with Brillance 64-slice computed tomography scanners. The image was reviewed by two or more otolaryngologists to identify the frontal recess cells. The nasofrontal isthmus diameter and the area of nasofrontal isthmus was reconstructed and measured with workstation. Then, we compared the radiological results of frontal recess cells with the frequency of frontal sinusitis. ResultsZZThe presence of anterior group of frontal recess cells showed no influence on the frontal recess anatomy. The presence of frontal bullar cell was significantly associated with the development of frontal sinusitis by simple (p=0.001) and multiple (p=0.038) logistic regression models. It was shown that the narrower the area of frontal isthmus the more developed were the frontal sinusitis, showing statistically significance in the simple (p=0.013) and multiple (p= 0.017) logistic regression models. ConclusionZZOur results also showed that similar results compared to previous Asianreport. The narrowness of nasofrontal isthmus could be the cause of frontal sinusitis. The frontal bullar cell could be the cause of frontal sinusitis encroaching on the frontal recess and affect the nasofrontal pathway.�

How can we use synergy to explain the value created by business analytics systems? In this article, we conceptualize and operationalize two important aspects of synergy: namely, the synergistic relationship and the synergistic outcome. We explore the enablers and mechanisms that are involved in a synergistic relationship between business analytics systems and customer relationship management systems and define it as the ability of systems to work together, span their boundaries and complement each other. Synergistic outcomes are the new business analytics–enabled customer relationship management systems that emerge from the synergistic relationship between business analytics systems and customer relationship management systems. Taking a whole system perspective, business analytics–enabled customer relationship management systems comprise the components and the emergent properties that arise from their interaction (e.g. the ability to cross-sell and up-sell based on advanced computational methods), in which the emergent properties are new because they do not exist in the individual components. We develop a research model that uses Synergistic Relationship and Synergistic Outcomes to explain the business value created by business analytics systems and customer relationship management systems, and we test this model using a survey of 201 managers in Australia and the United States. We find that the synergistic relationship plays a significant role in the creation of business analytics–enabled customer relationship management systems and subsequently business value. Business analytics–enabled customer relationship management systems—comprising business analytics systems, customer relationship management systems and their emergent properties—contribute to transactional, informational and strategic value. This goes beyond the value created by the business analytics and customer relationship management systems individually, as measured through statistical interaction.