Anemia is one of the most prevalent systemic complications in patients with cancer, substantially impairing quality of life and limiting the safe administration of cytoreductive therapies. Despite its clinical significance, the mechanisms by which tumors disrupt erythroid homeostasis remain incompletely understood. Here, we identify a previously unrecognized mechanism by which tumor-derived DNA directly drives cancer-associated anemia through pathological interaction with circulating red blood cells (RBCs). Specifically, we show that circulating tumor-derived DNA binds to lon peptidase 1 (LONP1), a mitochondrial protease aberrantly expressed on the surface of peripheral blood reticulocytes in tumor-bearing hosts. This interaction induces morphological alterations and apoptosis in reticulocytes, thereby triggering their premature clearance via erythrophagocytosis and contributing to anemia progression. Therapeutically, the enzymatic degradation of surface-bound DNA using Deoxyribonuclease I (DNase I) restores reticulocyte morphology, diminishes erythrophagocytic clearance, and alleviates anemia in tumor-bearing models. Moreover, combining DNase I with erythropoietin-driven stimulation of erythropoiesis produces synergistic hematologic improvement, simultaneously limiting pathological RBC clearance and enhancing RBC production. Together, these findings reveal a previously unappreciated DNA-mediated axis linking tumor burden to systemic erythroid dysfunction. This work establishes circulating tumor-derived DNA as an active pathogenic mediator in cancer-associated anemia and provides a mechanistically grounded combinatorial therapeutic strategy targeting both erythrocyte destruction and impaired erythropoiesis.

Saturated vapor pressure (Ps) is a crucial thermodynamic property in chemical engineering, playing an irreplaceable role in phase equilibrium calculations, separation process design, and mass transfer analysis. In this study, Atomic Connectivity Group Contribution (ACGC) and Spatial Group Contribution (SGC) models are developed for predicting Ps by introducing the normal boiling point (Tb). During the modeling process, the ACGC method primarily considers the topological structures of the molecules. Given the stereochemical characteristics of molecular configurations, the properties exhibited by the same functional group can vary depending on its spatial position or conformation within a molecule. The SGC method introducing the Spatial Group Factor (SGF) and Spatial Position Factor (SPF) is further applied to quantify the spatial structures of molecules. The lnPs(T, Tb)-ACGC and the lnPs(T, Tb)-SGC models achieve average absolute error (AAE) of 0.067 and 0.056 ln(kPa) for test set, respectively. Through internal validation, both models demonstrate excellent stability (Q2LOO: 0.9972 and 0.9981). The research results indicate that the lnPs(T, Tb)-SGC model shows better performance than the lnPs(T, Tb)-ACGC model in terms of Ps prediction accuracy. The lnPs(T, Tb)-ACGC model employs a strategy based on molecular shape factors and atomic connectivity factor, maintaining a relatively simple architecture while achieving efficient property estimation. In contrast, the lnPs(T, Tb)-SGC model demonstrates enhanced predictive performance through full consideration of spatial structural characteristics, albeit at the expense of increased model complexity.

Tumour-associated macrophages (TAMs) contribute to immune checkpoint blockade resistance, but their impact on intratumoural CD8⁺ T cell distribution remains unclear. Here we show that the expression of the glucose transporter SLC2A1 is spatially negatively correlated with CD8⁺ T cell distribution in both non-small-cell lung cancer (NSCLC) biopsies and murine tumour models. Tumour cell-specific Slc2a1 knockdown fails to reproduce the therapeutic benefit of SLC2A1 inhibition, whereas TAM-specific deletion of Slc2a1 suppresses tumour growth by enhancing the spatial homogeneity and effector function of intratumoural CD8⁺ T cells, thereby improving αPD-L1 efficacy. Spatial profiling of NSCLC specimens further revealed that SLC2A1⁺ TAM-enriched regions exhibit reduced CD8⁺ T cell density, and spatial proximity between these populations predicts resistance to αPD-(L)1 therapy. These findings identify SLC2A1⁺ TAMs as drivers of spatial CD8⁺ T cell exclusion and highlight TAM-specific SLC2A1 as a therapeutic target to overcome immune checkpoint blockade resistance in NSCLC. Wang, Chu, Chen, Wei and colleagues discover a subset of tumour-associated macrophages expressing SLC2A1 whose spatial proximity to CD8+ T cells drives resistance to anti-PD-L1 treatment in non-small-cell lung cancer.

A spatial group contribution (SGC) method is proposed for predicting the Delta vap H degrees (298.15 K), Delta sub H degrees (298.15 K), Delta fus H, and Delta fus S values of organic compounds. Relying solely on the count of groups to predict molecular properties in traditional group contribution (GC) largely overlooks the specificity conferred by the molecular stereochemistry. Our study introduces spatial group factor (SGF) and spatial position factor (SPF), designed to quantitatively characterize the three-dimensional morphological features of groups and their spatial position, respectively. The average absolute deviations of the isomers for four SGC models are reduced by 21.5, 23.3, 10.5, and 10.7%, respectively, compared with the traditional GC method, indicating that the proposed SGF and SPF facilitate the differentiation of isomers. The squared correlation coefficients (R test 2) of the test sets in all models are higher than 0.91, and their Q LOO 2 obtained by internal validation exceeds 0.89, indicating that all models have good predictive ability and stability.

Classical group contribution method, as one of the main methods for estimating thermodynamic properties, is developed with the number of groups, ignoring the influence of group characters. In this work, the spatial group contribution (SGC) method combining Euclidean distance and quantum properties is proposed, which uses the spatial group factor (SGF) and the spatial position factor (SPF) to reflect the spatial differences of the groups, thereby improving the limitations of the previous methods that only rely on topological structures. Five SGC models are established, including critical temperature ( T c ), critical pressure ( P c ), critical volume ( V c ), boiling point ( T b ), and melting point ( T m ), and the squared correlation coefficients ( R 2 training ) of 0.9935, 0.9925, 0.9988, 0.9828, and 0.8690 are obtained, respectively. After a series of rigorous validation procedures (external validation and internal validation), all models present excellent predictability ( R 2 test : 0.8690–0.9988) and stability ( Q 2 : 0.8344–0.9981). Compared with the atomic adjacent group (AAG) model, which is a traditional group contribution method, the absolute mean relative errors (AARE training ) of five models are reduced by 24.67%–69.26%. The position factor and spatial group factor crucially improve the models based on the number of groups. The spatiality‐based SGC method is of great significance for the prediction of thermodynamic properties and has the potential to be extended to more thermodynamic properties such as phase transition properties of enthalpy and entropy as well as saturated vapor pressure.

Rectal cancer, a prevalent malignant neoplasm within the digestive system, significantly jeopardizes patient health and quality of life. Accurate preoperative T-staging is critical for developing effective treatment strategies. In areas with limited medical resources, computed tomography (CT) has become the norm because of its popularity and economy and is an important method for the initial diagnosis of disease. Despite major advancements in computer vision in recent years, large-scale models have high demands on hardware and datasets, making them difficult to use and deploy in resource-limited environments. To address this challenge, we designed two lightweight modules, LightFire and ResLightFire, and developed a lightweight rectal cancer Tstaging network (LRCTNet). On this basis, we leveraged the swin transformer, transfer learning and knowledge distillation techniques to optimize the classification performance of the LRCTNet. The experimental results revealed that LRCTNet achieved a classification accuracy of 95.79%, precision of 93.91%, recall of 93.48%, F1 score of 93.70%, and Matthews correlation coefficient (MCC) of 94.38% while containing only 0.407 million parameters, which were much higher than those of lightweight models such as SqueezeNet, MobileNet, and EfficientNet. These results indicate that the model achieves a low misclassification rate and a low rate of missed detections, ensuring balanced performance in classification. The lightweight design of LRCTNet enables efficient deployment in resource-constrained environments without sacrificing accuracy, making it a valuable tool for rectal cancer diagnosis.

CD8+ T cell exclusion and dysfunction in the tumor microenvironment (TME) are among the most challenging obstacles for anti-PD-(L)1 therapy. Here, we report that tumor-infiltrating dendritic cell (DC)-specific expression of the deoxyribonuclease, DNASE1L3, is positively correlated with favorable outcomes of anti-PD-(L)1 treatment in cancer patients. DNASE1L3 conditional knockout in DCs leads to enhanced tumor growth and diminishes anti-PD-L1 therapeutic efficacy by impairing infiltration and effector functions of CD8+ T cells. Conversely, injection with DNASE1L3 promotes CD8+ T cell infiltration and reduces exhaustion in the TME, significantly retarding tumor growth and enhancing anti-PD-L1 response. DNASE1L3+ DCs can degrade neutrophil extracellular traps that suppress the spatial distribution of CD8+ T cells in tumors, enabling establishment of cytotoxic CD8+ T cell hubs in human cancers. Our findings reveal a role of DC in regulating intratumoral CD8+ T cells and identify DNASE1L3 as a promising target to improve anti-PD-(L)1 therapy.

Accurately predicting the density of organic compounds is essential in chemical engineering. This study develops a robust quantitative structure-property relationship (QSPR) model using a multiple linear regression (MLR) methodology, based on a comprehensive dataset of 5478 organic compounds and 23 866 data points to predict density over a broad temperature range (115.0 to 594.1 K). Notably, norm indices (NIs) are applied for QSPR modeling of organic compound density for the first time. The model demonstrates excellent predictive performance, with a squared correlation coefficient (R2) of 0.9953 and a mean absolute error (MAE) of 10.11 kg m-3. Rigorous internal, external, and extrapolation validations are applied to confirm the model's reliability, accuracy, and generalization. The model achieves an R2 value of 0.9951 and a MAE of 9.31 kg m-3 in external validation, while in internal validation using leave-one-out cross-validation, the corresponding values are 0.9951 and 10.51 kg m-3, respectively. Extrapolation validation, a novel approach recently introduced, further confirms the model's extrapolation ability, with most descriptors achieving the root mean square error (RMSE) of the test set (EV) values well below the training set's standard deviation (sigma 95 = 140.89 kg m-3), closely aligning with RMSEtest (model). The RMSE of forward test exhibits a significant increase for NI8 and NI27 when the extrapolation degree (ED) exceeds 0.02, which suggests that it is not recommended to apply these two NIs for extrapolation. Overall, the results validate the robustness and broad applicability of the rho(NI,T)-QSPR model, confirming its reliability for organic compound density prediction in industrial applications.

Vapor pressure (Ps) is a crucial thermodynamic property in the chemical industry, especially for the determination of the vapor-liquid equilibrium of the substance. In this work, the Antoine-based quantitative structure-property relationship for Ps is developed by introducing the easily measured normal boiling point (Tb) and the norm indices (NIs). During the modelling, the impact of strong interactions of groups on the Ps is considered by incorporating C = XN (X is O or S) and XH (X is O or S) into the new group-based NIs. An improved extrapolation degree analysis is performed for each NI, facilitating a more comprehensive assessment of the model's extrapolation capability as well as guiding the application of the model. The lnPs(NI,T,Tb)-QSPR model yields a squared correlation coefficient (R2) of 0.9975 and an average absolute error (AAE) of 0.076 ln(kPa). In contrast, the lnPs(NI,T)-QSPR model focusing solely on the NI achieves an R2 of 0.9770 and an AAE of 0.272 ln(kPa). Upon introducing the Tb, a notable reduction in the AAE for the lnPs(NI,T,Tb)-QSPR model is observed compared to the lnPs(NI,T)-QSPR model, with the AAE reduced by 72 %. Furthermore, the impact of Tb on the lnPs(NI,T,Tb)-QSPR model suggests that the AAE of the lnPs(NI,T,Tb)-QSPR model approaches that of the lnPs(NI,T)-QSPR model when the absolute error of Tb reaches 9 K. These results demonstrate that the introduction of the Tb can significantly enhance the prediction accuracy of Ps. The lnPs(NI,T,Tb)-QSPR model is a satisfactory and reliable method for estimating the Ps of substances.

The distinctive cavity structure of cyclodextrin, which results in binding properties, is credited with its application prospects in chemical, pharmacy, and material fields. The binding capacity can be regulated by substituting the hydroxyl groups on the cyclodextrins. It is possible to acquire anticipated binding properties by designing the modified groups on cyclodextrins. In this article, a data-driven model is proposed with a novel cyclodextrin/guest structure representation method to assist the cyclodextrin design. The model's performance is verified via several validations, as the squared correlation coefficients for cross-validation (Q2) and test set (R2test) are 0.801 and 0.841, respectively. With the proposed model and fluorescence experiments for cyclodextrin/bisphenol complexes, several cyclodextrin hosts, which have a strong binding capacity for bisphenols, are screened, synthesized, and characterized. The results show a controlled average absolute error of 0.605 M-1, suggesting the feasibility of data supplementation and molecular design. It is believed that the data-driven model can serve as theoretical assistance and a driving tool for the cyclodextrin complexes design, potentially leading to advancements in cyclodextrin's industrial applications and scientific research.

Objective To establish an animal model and evaluation system for lymph node metastasis of breast cancer. Methods A total of 60 female BALB/c mice (6~8 weeks old) were subjected, and then 6 models of lymph node metastasis (n=10) were constructed through injection at different parts in the mice with cell suspension of 4T1 breast cancer cells.Transgenic mice (n=5) of mouse mammary tumor virus-polyoma middle T antigen (MMTV-PyMT) were employed and served as model of spontaneous tumor metastasis.Then the advantages and disadvantages of different lymph node metastasis models were comprehensively evaluated from multiple aspects, such as operability, histomorphology and pathological detection, tumor growth rate and mouse survival. Results Among the 7 metastasis models, 4 models of lymph node metastasis were successfully established.Among them, the PyMT mouse spontaneous tumorigenesis model showed the best clinical reproduction, with a tumorigenesis rate of up to 100%, but had a disadvantage of poor experimental standardization.The hind paw-popliteal lymph node model had the fastest lymph node metastasis, easy operation and high repeatability, and a tumorigenesis rate of 100%, indicating its suitable for lymph node metastasis related research.The thigh subcutaneous-inguinal lymph node model also successfully simulated lymph node metastasis, with simple operation and high repeatability, a tumorigenesis rate of up to 100%, but its metastasis time was slightly longer than the hind paw-popliteal lymph node model.The inguinal lymph node-contralateral lymph node model was also a successful lymph node metastasis model, but with difficult operation, only 50% tumor-bearing rate, and poor repeatability.Lymph node metastasis model was not successfully established in the other 3 tumor-bearing models (under the tongue-internal jugular scapular tongue muscle lymph node model, bone marrow-inguinal lymph node model and right upper back skin-axillary lymph node model) in a short time, with no tumor cells observed in the pathological sections. Conclusion Through the comprehensive comparison of multiple models, mouse hind paw-popliteal lymph node model is the most suitable for conducting related research.

To overcome the limitations of empirical synthesis and expedite the discovery of new polymers, this work aims to develop a data-driven strategy for profoundly aiding in the design and screening of novel polyester materials. Initially, we collected 695 polyesters with their associated glass transition temperatures (Tgs) to develop a quantitative structure-property relationship (QSPR) model. The model underwent rigorous validation (external validation, internal validation, Y-random and application domain analysis) to demonstrate its robust predictive capabilities and high stability. Subsequently, by employing an in-silico retrosynthesis strategy, over 95000 virtual polyesters were designed, largely expanding the available space for polyester materials. External assessments highlight the good extrapolation ability of the QSPR model. Furthermore, we experimentally synthesized diverse virtual polyesters with Tgs covering a sufficient large temperature range. It is believed that this data-driven approach can drive future product development of polymer industry.

Immunotherapy is widely used in cancer treatment; however, only a subset of patients responds well to it. Significant efforts have been made to identify patients who will benefit from immunotherapy. Successful anti-tumor immunity depends on an intact cancer-immunity cycle, especially long-lasting CD8+ T-cell responses. Interferon (IFN)-α/β/IFN-γ/interleukin (IL)-15 pathways have been reported to be involved in the development of CD8+ T cells. And these pathways may predict responses to immunotherapy. Herein, we aimed to analyze multiple public databases to investigate whether IFN-α/β/IFN-γ/IL-15 pathways could be used to predict the response to immunotherapy. Results showed that IFN-α/β/IFN-γ/IL-15 pathways could efficiently predict immunotherapy response, and guanylate-binding protein 1 (GBP1) could represent the IFN-α/β/IFN-γ/IL-15 pathways. In public and private cohorts, we further demonstrated that GBP1 could efficiently predict the response to immunotherapy. Functionally, GBP1 was mainly expressed in macrophages and strongly correlated with chemokines involved in T-cell migration. Therefore, our study comprehensively investigated the potential role of GBP1 in immunotherapy, which could serve as a novel biomarker for immunotherapy and a target for drug development.

Objective To explore the sensitizing effect of manganese for radiotherapy against tumors and its possible mechanisms. Methods A total of 300 male C57BL/6 mice (6~8 weeks old, weighing 20~23 g) with subcutaneous tumor were randomly divided into 4 groups: control group, radiotherapy group, manganese treatment group, and combined radiotherapy and manganese treatment group. Nasal drip with 10 μg manganese adjuvant was applied to the mice from the latter 2 groups on day 9 of tumor bearing, then single dose of 20 Gy radiation was locally administered on day 10. Tumor growth and mouse survival were monitored regularly. The sensitizing effect of manganese on radiotherapy was determined by monitoring and comparing the tumor growth among different unilateral mouse models. Bilateral tumor-bearing model was used to examine the effect of manganese on abscopal effects induced by radiotherapy. Flow cytometry was used to illustrate the changes in tumor-infiltrating immune cells in unilateral tumor bearing model. Immunohistochemical staining was employed to evaluate the spleen function in unilateral tumor bearing mice. Results Based on repeated validation of 3 different unilateral tumor-bearing models, radiotherapy combined with manganese therapy significantly inhibited tumor growth and prolonged survival of mice (P < 0.05). The results of bilateral tumor-bearing model showed that manganese therapy enhanced abscopal effects of radiotherapy, and significant regression was observed in both side of tumor under radiotherapy or not (P < 0.05). Flow cytometry revealed that manganese further increased radiation-induced CD8+T infiltration (P < 0.05) and decreased radiation-induced infiltration of Treg cells and myeloid-derived suppressor cells (MDSCs) (P < 0.05). Furthermore, manganese increased lymphocyte reserve pool of the spleen and improved its function. Conclusion Manganese adjuvant could act as a sensitizing agent for radiotherapy, by improving the function of spleen and reprogramming the tumor microenvironment synergistically.

Background and Objective: Pulmonary sarcomatoid carcinoma (PSC) is a subset of non -small cell lung cancer (NSCLC) with highly malignant, aggressive, and heterogeneous features. Patients with this disease account for approximately 0.1-0.4% of lung cancer cases. The absence of comprehensive summaries on the basic biology and clinical treatments for PSC means there is limited systematic awareness and understanding of this rare disease. This paper provides an overview of the biological characteristics of PSC and systematically summarizes various treatment strategies available for patients with this disease. Methods: For this narrative review, we have searched literature related to the basic biology and clinical treatment approaches of PSC by searching the PubMed database for articles published from July 16, 1990 to August 29, 2023. The following keywords were used: "pulmonary sarcomatoid carcinoma", "genetic mutations", "immune microenvironment", "hypoxia", "angiogenesis", "overall survival", "surgery", "radiotherapy", "chemotherapy", and "immune checkpoint inhibitors". Key Content and Findings: Classical PSC comprises epithelial and sarcomatoid components, with most studies suggesting a common origin. PSC exhibits a higher tumor mutational burden (TMB) and mutation frequency than other types of NSCLC. The tumor microenvironment (TME) of PSC is characterized by hypoxia, hypermetabolism, elevated programmed cell death protein 1/programmed cell death-ligand 1 expression, and high immune cell infiltration. Treatment strategies for advanced PSC are mainly based on traditional NSCLC treatments, but PSC exhibits resistance to chemotherapy and radiotherapy. The advancement of genome sequencing has introduced targeted therapies as an option for mutation -positive PSC cases. Moreover, due to the characteristics of the immune microenvironment of PSC, many patients positively respond to immunotherapy, demonstrating its potential for the management of PSC. Conclusions: Although several studies have examined and assessed the TME of PSC, these are limited in quantity and quality, presenting challenges for research into the clinical treatment strategies for PSC. With the emergence of new technologies and the advancement of clinical research, for example, savolitinib's clinical study for MET exon 14 skipping mutations positive PSC patients have shown promising outcomes, more in-depth studies on PSC are eagerly anticipated.

Objective Current biomarkers for predicting immunotherapy response in non-small-cell lung cancer (NSCLC) are derived from invasive procedures with limited predictive accuracy. Thus, identifying a non-invasive predictive biomarker would improve patient stratification and precision immunotherapy.Methods and analysis In this retrospective multicohort study, the discovery cohort included 205 NSCLC patients screened from ORIENT-11 and an external validation (EV) cohort included 99 real-world NSCLC patients. The ‘onion-mode segmentation’ method was developed to extract ‘onion-mode perfusion’ (OMP) from contrast-enhanced CT images. The predictive performance of OMP or its combination with the PD-L1 Tumour Proportion Score (TPS) was evaluated by the area under the curve (AUC).Results High baseline OMP was associated with significantly longer survival and predicted patient response to combination anti-PD-(L)1 therapy in the discovery and EV cohorts. OMP complemented the PD-L1 TPS with superior predictive sensitivity (p=0.02). In the PD-L1 TPS<50% subgroup, OMP achieved an AUC of 0.77 for the estimation of treatment response (95% CI 0.66 to 0.86, p<0.0001). A simple bivariate model of OMP/PD-L1 robustly predicted therapeutic response in both the discovery (AUC 0.82, 95% CI 0.74 to 0.88, p<0.0001) and EV (AUC 0.80, 95% CI 0.67 to 0.89, p<0.0001) cohorts.Conclusion OMP, derived from routine CT examination, could serve as a non-invasive and cost-effective biomarker to predict NSCLC patient response to immune checkpoint inhibitor-based therapy. OMP could be used alone or in combination with other biomarkers to improve precision immunotherapy.

Background:Lymph node (LN) dissection is a common procedure for non-small cell lung cancer (NSCLC) to ascertain disease severity and treatment options. However, murine studies have indicated that excising tumor-draining LNs diminished immunotherapy effectiveness, though its applicability to clinical patients remains uncertain. Hence, the authors aim to illustrate the immunological implications of LN dissection by analyzing the impact of dissected LN (DLN) count on immunotherapy efficacy, and to propose a novel 'immunotherapy-driven' LN dissection strategy.Materials and methods:The authors conducted a retrospective analysis of NSCLC patients underwent anti-PD-1 immunotherapy for recurrence between 2018 and 2020, assessing outcomes based on DLN count stratification.Results:A total of 144 patients were included, of whom 59 had a DLN count less than or equal to 16 (median, IQR: 11, 7-13); 66 had a DLN count greater than 16 (median, IQR: 23, 19-29). With a median follow-up time of 14.3 months (95% CI: 11.0-17.6), the overall median progression-free survival (PFS) was 7.9 (95% CI: 4.1-11.7) months, 11.7 (95% CI: 7.9-15.6) months in the combination therapy subgroup, and 4.8 (95% CI: 3.1-6.4) months in the immunotherapy alone subgroup, respectively. In multivariable Cox analysis, DLN count less than or equal to 16 is associated with an improved PFS in all cohorts [primary cohort: HR=0.26 (95% CI: 0.07-0.89), P=0.03]; [validation cohort: HR=0.46 (95% CI: 0.22-0.96), P=0.04]; [entire cohort: HR=0.53 (95% CI: 0.32-0.89), P=0.02]. The prognostic benefit of DLN count less than or equal to 16 was more significant in immunotherapy alone, no adjuvant treatment, pN1, female, and squamous carcinoma subgroups. A higher level of CD8+ central memory T cell (Tcm) within LNs was associated with improved PFS (HR: 0.235, 95% CI: 0.065-0.845, P=0.027).Conclusions:An elevated DLN count (cutoff: 16) was associated with poorer immunotherapy efficacy in recurrent NSCLC, especially pronounced in the immunotherapy alone subgroup. CD8+Tcm proportions in LNs may also impact immunotherapy efficacy. Therefore, for patients planned for adjuvant immunotherapy, a precise rather than expanded lymphadenectomy strategy to preserve immune-depending LNs is recommended.

Universal property estimation for organics in chemical engineering poses a significant challenge. In this work, a consistent set of norm indices is employed for various thermodynamic properties, including critical properties (Pc, Vc and Tc), boiling points (Tb) and melting points (Tm). The Pc (6 4 3), Vc (6 3 7), Tc (8 8 6), Tb (5952) and Tm (7291) of diverse organic compounds have been utilized to establish these quantitative structure–property relationship models. The predictive ability of models is demonstrated with satisfactory results, with fitness coefficients of test sets (Rtest2) exceeding 0.969 for Pc, Vc, Tc and Tb. Internal validation indicates the robustness of these models with fitness coefficients (Q2) between 0.969 and 0.997. The performance of Tm model, in which Rtest2 is 0.837 and Q2 is 0.834, is considered reasonable and acceptable. The comparisons with the published models demonstrate our models exhibit high precision and diverse types of chemicals. In summary, these developed models with satisfactory performance and applicability, could be considered for practical applications in chemical and engineering and affirm the potential of norm indices for the universal property estimation.

The bioconcentration factor (BCF) is an indispensable parameter for evaluating the accumulation of chemicals in living organisms, and it has a significant role in determining the environmental risks of various chemicals. Nevertheless, due to many factors affecting BCF such as bio-absorption and metabolism, the accuracy of quantitative structure-activity relationship (QSAR) models for BCF is unsatisfactory. In this study, the four BCF-QSAR models based on multiple linear regression (MLR) and artificial neural network (ANN) algorithms have been developed with the help of the topological norm (TN) descriptor and the spatial norm (SN) descriptor (i.e., the atomic quantum chemical properties are considered as the third dimension of the molecular topological graph). After external validation to statistically validate the forecasting accuracy of the developed models, the ANN model based on the co-development of TN and SN showed high predictive accuracy, i.e., R2test of 0.9275. A high-accuracy QSAR model may provide reliable support for the precise evaluation of future aquatic risks and the setting of water quality standards. Subsequently, the four BCF-QSAR models are interpreted using SHapley Additive exPlanations. Besides, the rapid BCF estimation method and identifying important descriptors enable early alerts to monitor the potential aquatic pollutants.