Identification of Key Molecular Pathways and Associated Genes as Targets to Overcome Radiotherapy Resistance Using a Combination of Radiotherapy and Immunotherapy in Glioma Patients.

Recent mechanistic studies have indicated that combinations of radiotherapy (RT) plus immunotherapy (via CSF-1R inhibition) can serve as a strategy to overcome RT resistance and improve the survival of glioma mice. Given the high mortality rate for glioma, including low-grade glioma (LGG) patients, it is of critical importance to investigate the mechanism of the combination of RT and immunotherapy and further translate the mechanism from mouse studies to improve survival of RT-treated human glioma patients. Using the RNA-seq data from a glioma mouse study, 874 differentially expressed genes (DEGs) between the group of RT-treated mice at glioma recurrence and the group of mice with combination treatment (RT plus CSF-1R inhibition) were translated to the human genome to identify significant molecular pathways using the KEGG enrichment analysis. The enrichment analysis yields statistically significant signaling pathways, including the phosphoinositide 3-kinase (PI3K)/AKT pathway, Hippo pathway, and Notch pathway. Within each pathway, a candidate gene set was selected by Cox regression models as genetic biomarkers for resistance to RT and response to the combination of RT plus immunotherapies. Each Cox model is trained using a cohort of 295 RT-treated LGG patients from The Cancer Genome Atlas (TCGA) database and validated using a cohort of 127 RT-treated LGG patients from the Chinese Glioma Genome Atlas (CGGA) database. A four-DEG signature (ITGB8, COL9A3, TGFB2, JAG1) was identified from the significant genes within the three pathways and yielded the area under time-dependent ROC curve AUC = 0.86 for 5-year survival in the validation set, which indicates that the selected DEGs have strong prognostic value and are potential intervention targets for combination therapies. These findings may facilitate future trial designs for developing combination therapies for glioma patients.

Microfluidic formulation of food additives-loaded nanoparticles for antioxidation.

Browning has many important implications with nutrition and the shelf life of foods. Mitigating browning is of particular interest in food chemistry. The addition of antioxidants has been a common strategy to extend shelf life of drug and food products. In this work, we report a microfluidic technology for encapsulation of three common food additives (potassium metathionite (PMS), curcumin (CCM), and ß-carotene (ß-Car)) into nano-formulations using low-cost and readily available materials such as shellac. The food additives encapsulated nanoparticles provide a microenvironment that can prevent oxidation during daily storage. The results showed that the produced nanoparticles had a narrow size distribution with an average size of around 100 nm, were stable at conventional storage conditions (4 ºC) for 18 weeks, and had sustained release ability at 37 ºC, pH= 7.8, 160 rpm. In addition, further experiments showed that the formulation of hydrophobic additives, such as CCM and ß-Car did not only improve their bioavailability but also allowed for the encapsulation of a combination of ingredients. In addition, the antioxidants loaded nanoparticles demonstrated good biocompatibility, low toxicity to human cells. The longer release time of encapsulated food additives increases shelf life of foods and enhances consumer purchase preferences, which not only saves costs but also reduces waste. In summary, this study shows that such antioxidant-loaded nanoparticles provide a promising strategy in extending the shelf life of food products.

Sex-specific resting state brain network dynamics in patients with major depressive disorder.

Sex-specific neurobiological changes have been implicated in Major Depressive Disorder (MDD). Dysfunctions of the default mode network (DMN), salience network (SN) and frontoparietal network (FPN) are critical neural characteristics of MDD, however, the potential moderating role of sex on resting-state network dynamics in MDD has not been sufficiently evaluated. Thus, resting-state functional magnetic resonance imaging (fMRI) data were collected from 138 unmedicated patients with first-episode MDD (55 males) and 243 healthy controls (HCs; 106 males). Recurring functional network co-activation patterns (CAPs) were extracted, and time spent in each CAP (the total amount of volumes associated to a CAP), persistence (the average number of consecutive volumes linked to a CAP), and transitions across CAPs involving the SN, DMN and FPN were quantified. Relative to HCs, MDD patients exhibited greater persistence in a CAP involving activation of the DMN and deactivation of the FPN (DMN + FPN-). In addition, relative to the sex-matched HCs, the male MDD group spent more time in two CAPs involving the SN and DMN (i.e., DMN + SN- and DMN-SN + ) and transitioned more frequently from the DMN + FPN- CAP to the DMN + SN- CAP relative to the male HC group. Conversely, the female MDD group showed less persistence in the DMN + SN- CAP relative to the female HC group. Our findings highlight that the imbalance between SN and DMN could be a neurobiological marker supporting sex differences in MDD. Moreover, the dominance of the DMN accompanied by the deactivation of the FPN could be a sex-independent neurobiological correlate related to depression.

Ten-year follow-up of very-high risk hypertensive patients undergoing renal sympathetic denervation.

OBJECTIVES: Renal denervation (RDN) has been proven to be effective in lowering blood pressure (BP) in patients, but previous studies have had short follow-ups and have not examined the effects of RDN on major cardiovascular outcomes. This study aimed to demonstrate the effectiveness and safety of RDN in the long-term treatment of hypertension and to determine if it has an effect on cardiovascular outcomes. METHODS: All patients with resistant hypertension who underwent RDN between 2011 and 2015 at Tianjin First Central Hospital were included in the study. Patients were followed up at 1,5 and 10 years and the longest follow-up was 12 years. Data were collected on office BP, home BP, ambulatory BP monitoring (ABPM), renal function, antihypertensive drug regimen, major adverse events (including acute myocardial infarction, stroke, cardiovascular death and all cause death) and safety events. RESULTS: A total of 60 participants with mean age 50.37 ±â€Š15.19 years (43.33% female individuals) completed long-term follow-up investigations with a mean of 10.02 ±â€Š1.72 years post-RDN. Baseline office SBP and DBP were 179.08 ±â€Š22.05 and 101.17 ±â€Š16.57 mmHg under a mean number of 4.22 ±â€Š1.09 defined daily doses (DDD), with a reduction of -35.93/-14.76 mmHg as compared with baseline estimates ( P  < 0.0001). Compared with baseline, ambulatory SBP and DBP after 10-years follow-up were reduced by 14.31 ±â€Š10.18 ( P  < 0.001) and 9 ±â€Š4.35 ( P  < 0.001) mmHg, respectively. In comparison to baseline, participants were taking fewer antihypertensive medications ( P  < 0.001), and their mean heart rate had decreased ( P  < 0.001). Changes in renal function, as assessed by estimated glomerular filtration rate (eGFR) and creatinine, were within the expected rate of age-related decline. No major adverse events related to the RDN procedure were observed in long-term consequences. All-cause mortality and cardiovascular mortality rates were 10 and 8.34%, respectively, for the 10-year period. CONCLUSION: The BP-lowering effect of RDN was safely sustained for at least 10 years post-procedure. More importantly, to the best of my knowledge, this is the first study to explore cardiovascular and all-cause mortality at 10 years after RDN.

Multiscale modeling of drug resistance in glioblastoma with gene mutations and angiogenesis.

Drug resistance is a prominent impediment to the efficacy of targeted therapies across various cancer types, including glioblastoma (GBM). However, comprehending the intricate intracellular and extracellular mechanisms underlying drug resistance remains elusive. Empirical investigations have elucidated that genetic aberrations, such as gene mutations, along with microenvironmental adaptation, notably angiogenesis, act as pivotal drivers of tumor progression and drug resistance. Nonetheless, mathematical models frequently compartmentalize these factors in isolation. In this study, we present a multiscale agent-based model of GBM, encompassing cellular dynamics, intricate signaling pathways, gene mutations, angiogenesis, and therapeutic interventions. This integrative framework facilitates an exploration of the interplay between genetic mutations and the vascular microenvironment in shaping the dynamic evolution of tumors during treatment with tyrosine kinase inhibitor. Our simulations unveil that mutations influencing the migration and proliferation of tumor cells expedite the emergence of phenotype heterogeneity, thereby exacerbating tumor invasion under both treated and untreated conditions. Moreover, angiogenesis proximate to the tumor fosters a protumoral milieu, augmenting mutation-induced drug resistance by increasing the survival rate of tumor cells. Collectively, our findings underscore the dual roles of intrinsic genetic mutations and extrinsic microenvironmental adaptations in steering tumor growth and drug resistance. Finally, we substantiate our model predictions concerning the impact of gene mutations and angiogenesis on the responsiveness of targeted therapies by integrating single-cell RNA-seq, spatial transcriptomics, bulk RNA-seq, and clinical data from GBM patients. The multidimensional approach enhances our understanding of the complexities governing drug resistance in glioma and offers insights into potential therapeutic strategies.

ESICCC as a systematic computational framework for evaluation, selection, and integration of cell-cell communication inference methods.

Cell-cell communication (CCC) is critical for determining cell fates and functions in multicellular organisms. With the advent of single-cell RNA-sequencing (scRNA-seq) and spatial transcriptomics (ST), an increasing number of CCC inference methods have been developed. Nevertheless, a thorough comparison of their performances is yet to be conducted. To fill this gap, we developed a systematic benchmark framework called ESICCC to evaluate 18 ligand-receptor (LR) inference methods and five ligand/receptor-target inference methods using a total of 116 data sets, including 15 ST data sets, 15 sets of cell line perturbation data, two sets of cell type-specific expression/proteomics data, and 84 sets of sampled or unsampled scRNA-seq data. We evaluated and compared the agreement, accuracy, robustness, and usability of these methods. Regarding accuracy evaluation, RNAMagnet, CellChat, and scSeqComm emerge as the three best-performing methods for intercellular ligand-receptor inference based on scRNA-seq data, whereas stMLnet and HoloNet are the best methods for predicting ligand/receptor-target regulation using ST data. To facilitate the practical applications, we provide a decision-tree-style guideline for users to easily choose best tools for their specific research concerns in CCC inference, and develop an ensemble pipeline CCCbank that enables versatile combinations of methods and databases. Moreover, our comparative results also uncover several critical influential factors for CCC inference, such as prior interaction information, ligand-receptor scoring algorithm, intracellular signaling complexity, and spatial relationship, which may be considered in the future studies to advance the development of new methodologies.

Selective C(sp3)-S Bond Cleavage of Thioethers to Build Up Unsymmetrical Disulfides.

The selective C(sp3)-S bond cleavage of thioethers was first developed to prepare unsymmetrical disulfides by using electrophilic halogenation reagents. In this strategy, NBS (N-bromosuccinimide) achieves selective furfuryl C(sp3)-S bond cleavage of furfuryl alkylthioethers at room temperature. Meanwhile, NFSI (N-fluorobenzenesulfonimide) enables selective methyl C(sp3)-S bond cleavage of aryl and alkyl methylthioethers at an elevated temperature. Notably, the substrate scope investigation indicates that the order of selectivity of the C-S bond cleavage is furfuryl C(sp3)-S > benzyl C(sp3)-S > alkyl C(sp3)-S > C(sp2)-S bond. Moreover, this practical and operationally simple strategy also provides an important complementary way to access various unsymmetrical disulfides with excellent functional group tolerances and moderate to good yields.

Host-Guest Complexes of Pillar[5]arene as Components for Supramolecular Light-Harvesting Systems with Tunable Fluorescence.

A guest molecule containing a short alkyl spacer between the tetraphenylethylene group and the methylpyridinium group was designed and synthesized. After complexation with a water-soluble pillar[5]arene, the resulting host-guest complex can further self-assemble into fluorescence-emitting nanoparticles in water. By loading a commercially available dye Rhodamine 6G into the nanoparticles, an efficient artificial light-harvesting system with high donor/acceptor ratios (>400/1) was successfully constructed. The obtained systems show considerable antenna effects with values of more than 10 times. The system also exhibits tunable fluorescence emission behavior and can be used as a fluorescent ink for information encryption.

Multicellular network-informed survival model for identification of drug targets of gliomas.

Increasing evidence suggests that communication between tumor cells (TCs) and tumor-associated macrophages (TAMs) plays a substantial role in promoting progression of low-grade gliomas (LGG). Hence, it is becoming critical to model TAM-TC interplay and interrogate how the crosstalk affects prognosis of LGG patients. This paper proposed a translational research pipeline to construct the multicellular interaction gene network (MIGN) for identification of druggable targets to develop novel therapeutic strategies. Firstly, we selected immunotherapy-related feature genes (IFGs) for TAMs and TCs using RNA-seq data of glioma mice from preclinical trials. After translating the IFGs to human genome, we constructed TAM- and TC- associated networks separately, using a training set of 524 human LGGs. Subsequently, clustering analysis was performed within each network, and the concordance measure K-index was adopted to correlate gene clusters with patient survival. The MIGN was built by combining the clusters highly associated with survival in TAM- and TC-associated networks. We then developed a MIGN-based survival model to identify prognostic signatures comprised of ligands, receptors and hub genes. An independent cohort of 172 human LGG samples was leveraged to validate predictive accuracy of the signature. The areas under time-dependent ROC curves were 0.881, 0.867, and 0.839 with respect to 1-year, 3-year, and 5-year survival rates respectively in the validation set. Furthermore, literature survey was conducted on the signature genes, and potential clinical responses to targeted drugs were evaluated for LGG patients, further highlighting potential utilities of the MIGN signature to develop novel immunotherapies to extend survival of LGG patients.

Vehicle State Estimation Combining Physics-Informed Neural Network and Unscented Kalman Filtering on Manifolds.

This paper proposes a novel vehicle state estimation (VSE) method that combines a physics-informed neural network (PINN) and an unscented Kalman filter on manifolds (UKF-M). This VSE aimed to achieve inertial measurement unit (IMU) calibration and provide comprehensive information on the vehicle's dynamic state. The proposed method leverages a PINN to eliminate IMU drift by constraining the loss function with ordinary differential equations (ODEs). Then, the UKF-M is used to estimate the 3D attitude, velocity, and position of the vehicle more accurately using a six-degrees-of-freedom vehicle model. Experimental results demonstrate that the proposed PINN method can learn from multiple sensors and reduce the impact of sensor biases by constraining the ODEs without affecting the sensor characteristics. Compared to the UKF-M algorithm alone, our VSE can better estimate vehicle states. The proposed method has the potential to automatically reduce the impact of sensor drift during vehicle operation, making it more suitable for real-world applications.

Increased CaMKII activation and contrast changes of cardiac ß1-and ß3-Adrenergic signaling pathways in a humanized angiotensinogen model of hypertension.

Aims: Upregulation of Ca2+/calmodulin-dependent protein kinase II (CaMKII) contributes to the pathogenesis of cardiovascular disease, including hypertension. Transgenic rats expressing the human angiotensinogen gene [TGR (hAGT)L1623] are a new novel humanized model of hypertension that associates with declines in cardiac contractile function and ß-adrenergic receptor (AR) reserve. The molecular mechanisms are unclear. We tested the hypothesis that in TGR (hAGT)L1623 rats, left ventricular (LV) myocyte CaMKIIδ and ß3-AR are upregulated, but ß1-AR is down-regulated, which are important causes of cardiac dysfunction and ß-AR desensitization. Main methods: We compared LV myocyte CaMKIIδ, CaMKIIδ phosphorylation (at Thr287) (pCaMKIIδ), and ß1-and ß3-AR expressions and determined myocyte functional and [Ca2+]I transient ([Ca2+]iT) responses to ß-AR stimulation with and without pretreatment of myocytes using an inhibitor of CaMKII, KN-93 (10-6 M, 30 min) in male Sprague Dawley (SD; N = 10) control and TGR (hAGT)L1623 (N = 10) adult rats. Key findings: Hypertension in TGR (hAGT)L1623 rats was accompanied by significantly increased LV myocyte ß3-AR protein levels and reduced ß1-AR protein levels. CaMKIIδ phosphorylation (at Thr287), pCaMKIIδ was significantly increased by 35%. These changes were followed by significantly reduced basal cell contraction (dL/dtmax), relaxation (dR/dtmax), and [Ca2+]iT. Isoproterenol (10-8 M) produced significantly smaller increases in dL/dtmax, dR/dtmax, and [Ca2+]iT. Moreover, only in TGR (hAGT)L1623 rats, pretreatment of LV myocytes with KN-93 (10-6 M, 30 min) fully restored normal basal and isoproterenol-stimulated myocyte contraction, relaxation, and [Ca2+]iT. Significance: LV myocyte CaMKIIδ overactivation with associated contrast changes in ß3-AR and ß1-AR may be the key molecular mechanism for the abnormal contractile phenotype and ß-AR desensitization in this humanized model of hypertension.

Temperature-insensitive and fabrication-tolerant coarse wavelength division (de)multiplexing on a silica platform using an angled multimode interferometer.

Wavelength division (de)multiplexing (WDM) device is a crucial component for optical transmission networks. In this paper, we demonstrate a 4 channel WDM device with a 20 nm wavelength spacing on silica based planar lightwave circuits (PLC) platform. The device is designed using an angled multimode interferometer (AMMI) structure. Since there are fewer bending waveguides than other WDMs, the device footprint is smaller, at 21 mm × 0.4 mm. Owing to the low thermo-optic coefficient (TOC) of silica, a low temperature sensitivity of 10 pm/°C is achieved. The fabricated device exhibits high performance of an insertion loss (IL) lower than 1.6 dB, a polarization dependent loss (PDL) lower than 0.34 dB, and the crosstalk between adjacent channels lower than -19 dB. The 3 dB bandwidth is 12.3∼13.5 nm. Moreover, the device shows a high tolerance with a sensitivity of central wavelength to the width of multimode interferometer < 43.75 pm/nm.

On-Chip E00-E20 Mode Converter Based on Multi-Mode Interferometer.

Mode converters is a key component in mode-division multiplexing (MDM) systems, which plays a key role in signal processing and multi-mode conversion. In this paper, we propose an MMI-based mode converter on 2%-Δ silica PLC platform. The converter transfers E00 mode to E20 mode with high fabrication tolerance and large bandwidth. The experimental results show that the conversion efficiency can exceed -1.741 dB with the wavelength range of 1500 nm to 1600 nm. The measured conversion efficiency of the mode converter can reach -0.614 dB at 1550 nm. Moreover, the degradation of conversion efficiency is less than 0.713 dB under the deviation of multimode waveguide length and phase shifter width at 1550 nm. The proposed broadband mode converter with high fabrication tolerance is promising for on-chip optical network and commercial applications.

Trait- and state-like co-activation pattern dynamics in current and remitted major depressive disorder.

BACKGROUND: Distinguishing between trait- and state-like neural alternations in major depressive disorder (MDD) may advance our understanding of this recurring disorder. We aimed to investigate dynamic functional connectivity alternations in unmedicated individuals with current or past MDD using co-activation pattern analyses. METHODS: Resting-state functional magnetic resonance imaging data were acquired from individuals with first-episode current MDD (cMDD, n = 50), remitted MDD (rMDD, n = 44), and healthy controls (HCs, n = 64). Using a data-driven consensus clustering technique, four whole-brain states of spatial co-activation were identified and associated metrics (dominance, entries, transition frequency) were analyzed with respect to clinical characteristics. RESULTS: Relative to rMDD and HC, cMDD showed increased dominance and entries of state 1 (primarily involving default mode network (DMN)), and decreased dominance of state 4 (mostly involving frontal-parietal network (FPN)). Among cMDD, state 1 entries correlated positively with trait rumination. Conversely, relative to cMDD and HC, individuals with rMDD were characterized by increased state 4 entries. Relative to HC, both MDD groups showed increased state 4-to-1 (FPN to DMN) transition frequency but reduction in state 3 (spanning visual attention, somatosensory, limbic networks), with the former metric specifically related to trait rumination. LIMITATIONS: Further confirmation with longitudinal studies are required. CONCLUSIONS: Regardless of symptoms, MDD was characterized by increased FPN-to-DMN transitions and reduced dominance of a hybrid network. State-related effect emerged in regions critically implicated in repetitive introspection and cognitive control. Asymptomatic individuals with past MDD were uniquely linked to increased FPN entries. Our findings identify trait-like brain network dynamics that might increase vulnerability to future MDD.

Transition-metal-free C-S bond cleavage and transformation of organosulfur compounds.

The activation and transformation of organic chemical bonds is a fundamental scientific problem. In the past several decades, C-S bond cleavage for the construction of C-C and C-heteroatom bonds has received tremendous attention in organic chemistry. Although significant progress has been made in the field of transition metal strategies, a variety of novel transition-metal-free strategies have also been developed using halogenated reagents, oxidants, acids, and bases. Moreover, organic photochemical and electrochemical methods have also been developed to achieve transition-metal-free C-S bond cleavage of organosulfur compounds. To date, however, no comprehensive review of transition-metal-free strategies has been reported. Therefore, we herein provide a comprehensive review of the major advances in the field of the transition-metal-free C-S bond cleavage and transformation of organosulfur compounds, including thioethers, sulfoxides, sulfones, thioacetals, sulfonium salts, and sulfur ylides.

Structural abnormalities in adolescents with conduct disorder and high versus low callous unemotional traits.

There may be distinct conduct disorder (CD) etiologies and neural morphologies in adolescents with high callous unemotional (CU) traits versus low CU traits. Here, we employed surface-based morphometry methods to investigate morphological differences in adolescents diagnosed with CD [42 with high CU traits (CD-HCU) and 40 with low CU traits (CD-LCU)] and healthy controls (HCs, N = 115) in China. Whole-brain analyses revealed significantly increased cortical surface area (SA) in the left inferior temporal cortex and the right precuneus, but decreased SA in the left superior temporal cortex in the CD-LCU group, compared with the HC group. There were no significant cortical SA differences between the CD-HCU and the HC groups. Compared to the CD-HCU group, the CD-LCU group had a greater cortical thickness (CT) in the left rostral middle frontal cortex. Region-of-interest analyses revealed significant group differences in the right hippocampus, with CD-HCU group having lower right hippocampal volumes than HCs. We did not detect significant group differences in the amygdalar volume, however, the right amygdalar volume was found to be a significant moderator of the correlation between CU traits and the proactive aggression in CD patients. The present results suggested that the manifestations of CD differ between those with high CU traits versus low CU traits, and underscore the importance of sample characteristics in understanding the neural substrates of CD.

Persistent Ventral Anterior Cingulate Cortex and Resolved Amygdala Hyper-responses to Negative Outcomes After Depression Remission: A Combined Cross-sectional and Longitudinal Study.

BACKGROUND: Major depressive disorder (MDD) is a highly prevalent mood disorder affecting more than 300 million people worldwide. Biased processing of negative information and neural hyper-responses to negative events are hallmarks of depression. This study combined cross-sectional and longitudinal experiments to explore both persistent and resolved neural hyper-responses to negative outcomes from risky decision making in patients with current MDD (cMDD) and remitted MDD (rMDD). METHODS: A total of 264 subjects participated in the cross-sectional study, including 117 patients with medication-naïve, first-episode current depression; 45 patients with rMDD with only 1 episode of depression; and 102 healthy control subjects. Participants completed a modified balloon analog risk task during functional magnetic resonance imaging. In the longitudinal arm of the study, 42 patients with cMDD were followed and 26 patients with rMDD were studied again after 8 weeks of antidepressant treatment. RESULTS: Patients with cMDD showed hyper-responses to loss outcomes in multiple limbic regions including the amygdala and ventral anterior cingulate cortex (vACC). Amygdala but not vACC hyperactivity correlated with depression scores in patients with cMDD. Furthermore, amygdala hyperactivity resolved while vACC hyperactivity persisted in patients with rMDD in both cross-sectional and longitudinal studies. CONCLUSIONS: These findings provide consistent evidence supporting differential patterns of amygdala and vACC hyper-responses to negative outcomes during depression remission. Amygdala hyperactivity may be a symptomatic and state-dependent marker of depressive neural responses, while vACC hyperactivity may reflect a persistent and state-independent effect of depression on brain function. These findings offer new insights into the neural underpinnings of depression remission and prevention of depression recurrence.

Fluorescent Nanoassemblies in Water Exhibiting Tunable LCST Behavior and Responsive Light Harvesting Ability.

Responsive fluorescent nanomaterials have been received considerable attention in recent years. In this work, a bola-type amphiphilic molecule, CSO, was synthesized which contains a hydrophobic cyanostilbene core and hydrophilic oligo(ethylene glycol) (OEG) coils at both sides. The cyanostilbene group is aggregation-induced emission (AIE) active, while the OEG coils are thermo-responsive. As a result, the CSO molecules can self-assemble into blue-fluorescent nanoassemblies with lower critical solution temperature (LCST) behavior in aqueous media. It is noteworthy that the LCST behavior can be reversibly regulated with changes in concentration and the introduction of K+ . Intriguingly, fluorescence of CSO assembly shows a blue-shift upon heating. Finally, by employing CSO as a light capturing antenna and energy donor, an artificial light harvesting system with tunable emission and thermo-responsive characteristics was fabricated. This study not only demonstrates an integrated approach to create responsive fluorescent nanomaterials, but also shows great potential for producing luminescent materials and mimicking photosynthesis in nature.

Benchmarking and integration of methods for deconvoluting spatial transcriptomic data.

MOTIVATION: The rapid development of spatial transcriptomics (ST) approaches has provided new insights into understanding tissue architecture and function. However, the gene expressions measured at a spot may contain contributions from multiple cells due to the low-resolution of current ST technologies. Although many computational methods have been developed to disentangle discrete cell types from spatial mixtures, the community lacks a thorough evaluation of the performance of those deconvolution methods. RESULTS: Here, we present a comprehensive benchmarking of 14 deconvolution methods on four datasets. Furthermore, we investigate the robustness of different methods to sequencing depth, spot size and the choice of normalization. Moreover, we propose a new ensemble learning-based deconvolution method (EnDecon) by integrating multiple individual methods for more accurate deconvolution. The major new findings include: (i) cell2loction, RCTD and spatialDWLS are more accurate than other ST deconvolution methods, based on the evaluation of three metrics: RMSE, PCC and JSD; (ii) cell2location and spatialDWLS are more robust to the variation of sequencing depth than RCTD; (iii) the accuracy of the existing methods tends to decrease as the spot size becomes smaller; (iv) most deconvolution methods perform best when they normalize ST data using the method described in their original papers; and (v) the integrative method, EnDecon, could achieve more accurate ST deconvolution. Our study provides valuable information and guideline for practically applying ST deconvolution tools and developing new and more effective methods. AVAILABILITY AND IMPLEMENTATION: The benchmarking pipeline is available at https://github.com/SunXQlab/ST-deconvoulution. An R package for EnDecon is available at https://github.com/SunXQlab/EnDecon. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.