Revealing the algicidal characteristics of Maribacter dokdonensis: An investigation into bacterial strain P4 isolated from Karenia mikimotoi bloom water.

Harmful algal blooms (HABs) are a global environmental concern, causing significant economic losses in fisheries and posing risks to human health. Algicidal bacteria have been suggested as a potential solution to control HABs, but their algicidal efficacy is influenced by various factors. This study aimed to characterize a novel algicidal bacterium, Maribacter dokdonensis (P4), isolated from a Karenia mikimotoi (Hong Kong strain, KMHK) HAB and assess the impact of P4 and KMHK's doses, growth phase, and algicidal mode and the axenicity of KMHK on P4's algicidal effect. Our results demonstrated that the algicidal effect of P4 was dose-dependent, with the highest efficacy at a dose of 25% v/v. The study also determined that P4's algicidal effect was indirect, with the P4 culture and the supernatant, but not the bacterial cells, showing significant effects. The algicidal efficacy was higher when both P4 and KMHK were in the stationary phase. Furthermore, the P4 culture at the log phase could effectively kill KMHK cells at the stationary phase, with higher algicidal efficacy in the bacterial culture than that of the supernatant alone. Interestingly, P4's algicidal efficacy was significantly higher when co-culturing with xenic KMHK (~90% efficacy at day 1) than that with the axenic KMHK (~50% efficacy at day 1), suggesting the presence of other bacteria could regulate P4's algicidal effect. The bacterial strain P4 also exhibited remarkable algicidal efficacy on four other dinoflagellate species, particularly the armored species. These results provide valuable insights into the algicidal effect of M. dokdonensis on K. mikimotoi and on their interactions.

The Hessian fly resistance gene HvRHF1 is localized in an NBS-LRR gene cluster in barley.

Hessian fly (Mayetiola destructor Say) is a significant pest in cereal crops, causing substantial yield losses worldwide. While host resistance is the most efficient method for pest control, research on genetic characterization of Hessian fly resistance in barley (Hordeum vulgare L.) has been limited, and the underlying resistance mechanism remains largely unknown. In this study, we conducted fine mapping of a crucial Hessian fly resistance locus, known as HvRHF1, using a biparental population. Assisted with genetic markers and robust phenotyping assay, we pinpointed the HvRHF1 gene to an ~ 82 kb region on chromosome 4H. Gene prediction and annotation revealed that the HvRHF1 locus comprises three complete NBS-LRR genes, which are characteristic of disease resistance genes. As a result, our study not only provides valuable resources for resistance in barley and genetic tools for breeding, but also identifies candidate genes that lay the foundation for cloning HvRHF1. This endeavor will significantly contribute to our understanding of the molecular mechanisms underlying cereal resistance to Hessian fly.

MESBC: A novel mutually exclusive spectral biclustering method for cancer subtyping.

Many soft biclustering algorithms have been developed and applied to various biological and biomedical data analyses. However, few mutually exclusive (hard) biclustering algorithms have been proposed, which could better identify disease or molecular subtypes with survival significance based on genomic or transcriptomic data. In this study, we developed a novel mutually exclusive spectral biclustering (MESBC) algorithm based on spectral method to detect mutually exclusive biclusters. MESBC simultaneously detects relevant features (genes) and corresponding conditions (patients) subgroups and, therefore, automatically uses the signature features for each subtype to perform the clustering. Extensive simulations revealed that MESBC provided superior accuracy in detecting pre-specified biclusters compared with the non-negative matrix factorization (NMF) and Dhillon's algorithm, particularly in very noisy data. Further analysis of the algorithm on real datasets obtained from the TCGA database showed that MESBC provided more accurate (i.e., smaller p-value) overall survival prediction in patients with lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC) cancers when compared to the existing, gold-standard subtypes for lung cancers (integrative clustering). Furthermore, MESBC detected several genes with significant prognostic value in both LUAD and LUSC patients. External validation on an independent, unseen GEO dataset of LUAD showed that MESBC-derived clusters based on TCGA data still exhibited clear biclustering patterns and consistent, outstanding prognostic predictability, demonstrating robust generalizability of MESBC. Therefore, MESBC could potentially be used as a risk stratification tool to optimize the treatment for the patient, improve the selection of patients for clinical trials, and contribute to the development of novel therapeutic agents.

Image-Based Subtype Classification for Glioblastoma Using Deep Learning: Prognostic Significance and Biologic Relevance.

PURPOSE: To apply deep learning algorithms to histopathology images, construct image-based subtypes independent of known clinical and molecular classifications for glioblastoma, and produce novel insights into molecular and immune characteristics of the glioblastoma tumor microenvironment. MATERIALS AND METHODS: Using whole-slide hematoxylin and eosin images from 214 patients with glioblastoma in The Cancer Genome Atlas (TCGA), a fine-tuned convolutional neural network model extracted deep learning features. Biclustering was used to identify subtypes and image feature modules. Prognostic value of image subtypes was assessed via Cox regression on survival outcomes and validated with 189 samples from the Clinical Proteomic Tumor Analysis Consortium (CPTAC) data set. Morphological, molecular, and immune characteristics of glioblastoma image subtypes were analyzed. RESULTS: Four distinct subtypes and modules (imClust1-4) were identified for the TCGA patients with glioblastoma on the basis of the image feature data. The glioblastoma image subtypes were significantly associated with overall survival (OS; P = .028) and progression-free survival (P = .003). Apparent association was also observed for disease-specific survival (P = .096). imClust2 had the best prognosis for all three survival end points (eg, after 25 months, imClust2 had >7% surviving patients than the other subtypes). Examination of OS in the external validation using the unseen CPTAC data set showed consistent patterns. Multivariable Cox analyses confirmed that the image subtypes carry unique prognostic information independent of known clinical and molecular predictors. Molecular and immune profiling revealed distinct immune compositions of the tumor microenvironment in different image subtypes and may provide biologic explanations for the patterns in patients' outcomes. CONCLUSION: Our image-based subtype classification on the basis of deep learning models is a novel tool to refine risk stratification in cancers. The image subtypes detected for glioblastoma represent a promising prognostic biomarker with distinct molecular and immune characteristics and may facilitate developing novel, individualized immunotherapies for glioblastoma.

Identification of Sr67, a new gene for stem rust resistance in KU168-2 located close to the Sr13 locus in wheat.

KEY MESSAGE: Sr67 is a new stem rust resistance gene that represents a new resource for breeding stem rust resistant wheat cultivars Re-appearance of stem rust disease, caused by the fungal pathogen Puccinia graminis f. sp. tritici (Pgt), in different parts of Europe emphasized the need to develop wheat varieties with effective resistance to local Pgt populations and exotic threats. A Kyoto University wheat (Triticum aestivum L.) accession KU168-2 was reported to carry good resistance to leaf and stem rust. To identify the genomic region associated with the KU168-2 stem rust resistance, a genetic study was conducted using a doubled haploid (DH) population from the cross RL6071 × KU168-2. The DH population was phenotyped with three Pgt races (TTKSK, TPMKC, and QTHSF) and genotyped using the Illumina 90 K wheat SNP array. Linkage mapping showed the resistance to all three Pgt races was conferred by a single stem rust resistance (Sr) gene on chromosome arm 6AL, associated with Sr13. Presently, four Sr13 resistance alleles have been reported. Sr13 allele-specific KASP and STARP markers, and sequencing markers all showed null alleles in KU168-2. KU168-2 showed a unique combination of seedling infection types for five Pgt races (TTKSK, QTHSF, RCRSF, TMRTF, and TPMKC) compared to Sr13 alleles. The phenotypic uniqueness of the stem rust resistance gene in KU168-2 and null alleles for Sr13 allele-specific markers showed the resistance was conferred by a new gene, designated Sr67. Since Sr13 is less effective in hexaploid background, Sr67 will be a good source of stem rust resistance in bread wheat breeding programs.

Evaluating Prognostic Value of Dynamics of Circulating Lactate Dehydrogenase in Colorectal Cancer Using Modeling and Machine Learning.

Pretreatment serum lactate dehydrogenase (LDH) levels have been associated with poor prognosis in several types of cancer, including metastatic colorectal cancer (mCRC). However, very few models link survival to longitudinal LDH measured repeatedly over time during treatment. We investigated the prognostic value of on-treatment LDH dynamics in mCRC. Using data from two large phase III studies (2L and 3L+ mCRC settings, n = 824 and 210, respectively), we found that integrating longitudinal LDH data with baseline risk factors significantly improved survival prediction. Current LDH values performed best, enhancing discrimination ability (area under the receiver operating characteristic curve) by 4.5~15.4% and prediction accuracy (Brier score) by 3.9~15.0% compared with baseline variables. Combining all longitudinal LDH markers further improved predictive performance. After controlling for baseline covariates and other longitudinal LDH indicators, current LDH levels remained a significant risk factor in mCRC, increasing mortality risk by over 90% (P < 0.001) in 2L patients and 60-70% (P < 0.01) in 3L+ patients per unit increment in current log (LDH). Machine-learning techniques, like functional principal component analysis (FPCA), extracted informative features from longitudinal LDH data, capturing over 99% of variability and allowing prediction of survival. Unsupervised clustering based on the extracted FPCA features stratified patients into three groups with distinct LDH dynamics and survival outcomes. Hence, our approaches offer a valuable and cost-effective way for risk stratification and improves survival prediction in mCRC using LDH trajectories.

GGNpTCR: A Generative Graph Structure Neural Network for Predicting Immunogenic Peptides for T-cell Immune Response.

Identifying the interactions between T-cell receptor (TCRs) and human antigens is a crucial step in developing new vaccines, diagnostics, and immunotherapy. Current methods primarily focus on learning binding patterns from known TCR binding repertoires by using sequence information alone without considering the binding specificity of new antigens or exogenous peptides that have not appeared in the training set. Furthermore, the spatial structure of antigens plays a critical role in immune studies and immunotherapy, which should be addressed properly in the identification of interacting TCR-antigen pairs. In this study, we introduced a novel deep learning framework based on generative graph structures, GGNpTCR, for predicting interactions between TCR and peptides from sequence information. Results of real data analysis indicate that our model achieved excellent prediction for new antigens unseen in the training data set, making significant improvements compared to existing methods. We also applied the model to a large COVID-19 data set with no antigens in the training data set, and the improvement was also significant. Furthermore, through incorporation of additional supervised mechanisms, GGNpTCR demonstrated the ability to precisely forecast the locations of peptide-TCR interactions within 3D configurations. This enhancement substantially improved the model's interpretability. In summary, based on the performance on multiple data sets, GGNpTCR has made significant progress in terms of performance, universality, and interpretability.

Identification of robust yield quantitative trait loci derived from cultivated emmer for durum wheat improvement.

Durum wheat (Triticum turgidum ssp. durum L.) is an important world food crop used to make pasta products. Compared to bread wheat (Triticum aestivum L.), fewer studies have been conducted to identify genetic loci governing yield-component traits in durum wheat. A potential source of diversity for durum is its immediate progenitor, cultivated emmer (T. turgidum ssp. dicoccum). We evaluated two biparental populations of recombinant inbred lines (RILs) derived from crosses between the durum lines Ben and Rusty and the cultivated emmer wheat accessions PI 41025 and PI 193883, referred to as the Ben × PI 41025 (BP025) and Rusty × PI 193883 (RP883) RIL populations, respectively. Both populations were evaluated under field conditions in three seasons with an aim to identify quantitative trait loci (QTLs) associated with yield components and seed morphology that were expressed in multiple environments. A total of 44 and 34 multi-environment QTLs were identified in the BP025 and RP883 populations, respectively. As expected, genetic loci known to govern domestication and development were associated with some of the QTLs, but novel QTLs derived from the cultivated emmer parents and associated with yield components including spikelet number, grain weight, and grain size were identified. These QTLs offer new target loci for durum wheat improvement, and toward that goal, we identified five RILs with increased grain weight and size compared to the durum parents. These materials along with the knowledge of stable QTLs and associated markers can help to expedite the development of superior durum varieties.

A Novel and Efficient Digital Pathology Classifier for Predicting Cancer Biomarkers Using Sequencer Architecture.

In digital pathology tasks, transformers have achieved state-of-the-art results, surpassing convolutional neural networks (CNNs). However, transformers are usually complex and resource intensive. This study developed a novel and efficient digital pathology classifier called DPSeq to predict cancer biomarkers through fine-tuning a sequencer architecture integrating horizontal and vertical bidirectional long short-term memory networks. Using hematoxylin and eosin-stained histopathologic images of colorectal cancer from two international data sets (The Cancer Genome Atlas and Molecular and Cellular Oncology), the predictive performance of DPSeq was evaluated in a series of experiments. DPSeq demonstrated exceptional performance for predicting key biomarkers in colorectal cancer (microsatellite instability status, hypermutation, CpG island methylator phenotype status, BRAF mutation, TP53 mutation, and chromosomal instability), outperforming most published state-of-the-art classifiers in a within-cohort internal validation and a cross-cohort external validation. In addition, under the same experimental conditions using the same set of training and testing data sets, DPSeq surpassed four CNNs (ResNet18, ResNet50, MobileNetV2, and EfficientNet) and two transformer (Vision Transformer and Swin Transformer) models, achieving the highest area under the receiver operating characteristic curve and area under the precision-recall curve values in predicting microsatellite instability status, BRAF mutation, and CpG island methylator phenotype status. Furthermore, DPSeq required less time for both training and prediction because of its simple architecture. Therefore, DPSeq appears to be the preferred choice over transformer and CNN models for predicting cancer biomarkers.

Linkage drag analysis in three Aegilops speltoides introgressions carrying Sr47 in modern durum and hard red spring wheat germplasm.

KEY MESSAGE: Yield and quality tests of wheat lines derived from RWG35 show they carry little, or no linkage drag and are the preferred source of Sr47 for stem rust resistance. Three durum wheat (Triticum turgidum L. subsp. durum) lines, RWG35, RWG36, and RWG37 carrying slightly different Aegilops speltoides introgressions, but each carrying the Sr47 stem rust resistance gene, were backcrossed to three durum and three hard red spring (HRS) wheat (Triticum aestivum L.) cultivars to produce 18 backcross populations. Each population was backcrossed to the recurrent parent six times and prepared for yield trials to test for linkage drag. Lines carrying the introgression (S-lines) were compared to euploid sibling lines (W-lines) and their parent. Yield trials were conducted from 2018 to 2021 at three locations. Three agronomic and several quality traits were studied. In durum, lines derived from RWG35 had little or no linkage drag. Lines derived from RWG36 and RWG37 still retained linkage drag, most notably involving yield and thousand kernel weight, but also test weight, falling number, kernel hardness index, semolina extract, semolina protein content, semolina brightness, and peak height. In HRS wheat, the results were more complex, though the general result of RWG35 lines having little or no linkage drag and RWG36 and RWG37 lines retaining linkage drag still applied. But there was heterogeneity in the Glenn35S lines, and Linkert lines had problems combining with the Ae. speltoides introgressions. We concluded that introgressions derived from RWG35 either had eliminated linkage drag or any negative effects were minor in nature. We recommend that breeders who wish to incorporate Sr47 into their cultivars should work exclusively with germplasm derived from RWG35.

Waist-to-Height Ratio Is a Stronger Mediator in the Association between DASH Diet and Hypertension: Potential Micro/Macro Nutrients Intake Pathways.

Several studies have demonstrated that adhering to the Dietary Approaches to Stop Hypertension (DASH) diet may result in decreased blood pressure levels and hypertension risk. This may be an effect of a reduction in central obesity. In the current study, we explored the mediation role of multiple anthropometric measurements in association with DASH score and hypertension risk, and we investigated potential common micro/macro nutrients that react with the obesity-reduction mechanism. Our study used data from the National Health and Nutrition Examination Survey (NHANES). Important demographic variables, such as gender, race, age, marital status, education attainment, poverty income ratio, and lifestyle habits such as smoking, alcohol drinking, and physical activity were collected. Various anthropometric measurements, including weight, waist circumference, body mass index (BMI), and waist-to-height ratio (WHtR) were also obtained from the official website. The nutrient intake of 8224 adults was quantified through a combination of interviews and laboratory tests. We conducted stepwise regression to filter the most important anthropometric measurements and performed a multiple mediation analysis to test whether the selected anthropometric measurements had mediation effects on the total effect of the DASH diet on hypertension. Random forest models were conducted to identify nutrient subsets associated with the DASH score and anthropometric measurements. Finally, associations between common nutrients and DASH score, anthropometric measurements, and risk of hypertension were respectively evaluated by a logistic regression model adjusting for possible confounders. Our study revealed that BMI and WHtR acted as full mediators between DASH score and high blood pressure levels. Together, they accounted for more than 45% of the variation in hypertension. Interestingly, WHtR was found to be the strongest mediator, explaining approximate 80% of the mediating effect. Furthermore, we identified a group of three commonly consumed nutrients (sodium, potassium, and octadecatrienoic acid) that had opposing effects on DASH score and anthropometric measurements. These nutrients were also found to be associated with hypertension in the same way as BMI and WHtR in univariate regression models. The most important among these nutrients was sodium, which was negatively correlated with the DASH score (ß = -0.53, 95% CI = -0.56~-0.50, p < 0.001) and had a positive association with BMI (ß = 0.04, 95% CI = 0.01~0.07, p = 0.02), WHtR (ß = 0.06, 95% CI = 0.03~0.09, p < 0.001), and hypertension (OR = 1.09, 95% CI = 1.01~1.19, p = 0.037). Our investigation revealed that the WHtR exerts a greater mediating effect than BMI on the correlation between the DASH diet and hypertension. Notably, we identified a plausible nutrient intake pathway involving sodium, potassium, and octadecatrienoic acid. Our findings suggested that lifestyle modifications that emphasize the reduction of central obesity and the attainment of a well-balanced micro/macro nutrient profile, such as the DASH diet, could potentially be efficacious in managing hypertension.

CD142 Identifies Neoplastic Desmoid Tumor Cells, Uncovering Interactions Between Neoplastic and Stromal Cells That Drive Proliferation.

The interaction between neoplastic and stromal cells within a tumor mass plays an important role in cancer biology. However, it is challenging to distinguish between tumor and stromal cells in mesenchymal tumors because lineage-specific cell surface markers typically used in other cancers do not distinguish between the different cell subpopulations. Desmoid tumors consist of mesenchymal fibroblast-like cells driven by mutations stabilizing beta-catenin. Here we aimed to identify surface markers that can distinguish mutant cells from stromal cells to study tumor-stroma interactions. We analyzed colonies derived from single cells from human desmoid tumors using a high-throughput surface antigen screen, to characterize the mutant and nonmutant cells. We found that CD142 is highly expressed by the mutant cell populations and correlates with beta-catenin activity. CD142-based cell sorting isolated the mutant population from heterogeneous samples, including one where no mutation was previously detected by traditional Sanger sequencing. We then studied the secretome of mutant and nonmutant fibroblastic cells. PTX3 is one stroma-derived secreted factor that increases mutant cell proliferation via STAT6 activation. These data demonstrate a sensitive method to quantify and distinguish neoplastic from stromal cells in mesenchymal tumors. It identifies proteins secreted by nonmutant cells that regulate mutant cell proliferation that could be therapeutically. Significance: Distinguishing between neoplastic (tumor) and non-neoplastic (stromal) cells within mesenchymal tumors is particularly challenging, because lineage-specific cell surface markers typically used in other cancers do not differentiate between the different cell subpopulations. Here, we developed a strategy combining clonal expansion with surface proteome profiling to identify markers for quantifying and isolating mutant and nonmutant cell subpopulations in desmoid tumors, and to study their interactions via soluble factors.

Intra- and inter-habitat variation in sediment heavy metals, antibiotics and ecological risks in Mai Po RAMSAR, China.

Distribution of heavy metals (HMs) and antibiotics (ABs) in surface sediments of three habitats: mudflat, mangrove and gei wai (inter-tidal shrimp ponds), at Mai Po RAMSAR were determined with inductively coupled plasma and liquid chromatograph tandem - mass spectrometry, respectively. Eight HMs (Cr, As, Pb, Cd, Mn, Ni, Cu and Zn), and ten ABs (tetracyclines, quinolones, macrolides and sulphonamides) were detected in all habitats, with relatively lower concentration in gei wai. Ecological risk assessment based on PNEC revealed that HMs posed a higher ecological risk to microorganisms than ABs. All metals except Mn were above their respective threshold effect levels according to sediment quality guidelines, indicating their potential toxicity to benthos. The enrichment factor and geo-accumulation index on background values suggested sediments were moderately polluted by Zn, Cu and Cd, possibly from anthropogenic inputs. This study implies that HMs pollution must be prevented through proper regulation of agricultural and industrial discharge.

Association Mapping of Resistance to Tan Spot in the Global Durum Panel.

Tan spot, caused by the necrotrophic fungal pathogen Pyrenophora tritici-repentis (Ptr), is an important disease of durum and common wheat worldwide. Compared with common wheat, less is known about the genetics and molecular basis of tan spot resistance in durum wheat. We evaluated 510 durum lines from the Global Durum Wheat Panel (GDP) for sensitivity to the necrotrophic effectors (NEs) Ptr ToxA and Ptr ToxB and for reaction to Ptr isolates representing races 1 to 5. Overall, susceptible durum lines were most prevalent in South Asia, the Middle East, and North Africa. Genome-wide association analysis showed that the resistance locus Tsr7 was significantly associated with tan spot caused by races 2 and 3, but not races 1, 4, or 5. The NE sensitivity genes Tsc1 and Tsc2 were associated with susceptibility to Ptr ToxC- and Ptr ToxB-producing isolates, respectively, but Tsn1 was not associated with tan spot caused by Ptr ToxA-producing isolates, which further validates that the Tsn1-Ptr ToxA interaction does not play a significant role in tan spot development in durum. A unique locus on chromosome arm 2AS was associated with tan spot caused by race 4, a race once considered avirulent. A novel trait characterized by expanding chlorosis leading to increased disease severity caused by the Ptr ToxB-producing race 5 isolate DW5 was identified, and this trait was governed by a locus on chromosome 5B. We recommend that durum breeders select resistance alleles at the Tsr7, Tsc1, Tsc2, and the chromosome 2AS loci to obtain broad resistance to tan spot.

The wheat stem rust resistance gene Sr43 encodes an unusual protein kinase.

To safeguard bread wheat against pests and diseases, breeders have introduced over 200 resistance genes into its genome, thus nearly doubling the number of designated resistance genes in the wheat gene pool1. Isolating these genes facilitates their fast-tracking in breeding programs and incorporation into polygene stacks for more durable resistance. We cloned the stem rust resistance gene Sr43, which was crossed into bread wheat from the wild grass Thinopyrum elongatum2,3. Sr43 encodes an active protein kinase fused to two domains of unknown function. The gene, which is unique to the Triticeae, appears to have arisen through a gene fusion event 6.7 to 11.6 million years ago. Transgenic expression of Sr43 in wheat conferred high levels of resistance to a wide range of isolates of the pathogen causing stem rust, highlighting the potential value of Sr43 in resistance breeding and engineering.

Clearance as an Early Indicator of Efficacy for Therapeutic Monoclonal Antibodies: Circumventing Dose Selection Challenges in Oncology.

BACKGROUND AND OBJECTIVE: The designs of first-in-human (FIH) studies in oncology (e.g., 3 + 3 dose escalation design) usually do not provide a sufficient sample size to determine the dose-response relationship for efficacy. This study aimed to assess the feasibility of using monoclonal antibody (mAb) clearance as a biomarker for efficacy to facilitate the identification of potentially efficacious doses across cancer types and drug targets. METHODS: We performed electronic searches of the Drugs@FDA website, the European Medicines Agency website, and PubMed to identify reports of FIH trials of approved mAbs in oncology. The clearance, half-life, and overall response rate (ORR) data for the mAbs at different dose levels were extracted. RESULTS: Twenty-five approved mAbs were included in this study. As expected, due to the small sample sizes in FIH studies, there was no clear dose-response for ORR. However, we found a clear negative association between mAb clearance and ORR across tumors/drug targets, and a clear negative dose-clearance relationship, with clearance decreasing and saturated at high dose levels. The approved mAb doses (1-25 mg/kg) are approximately 2-fold the saturation doses (1-10 mg/kg). The associated clearance values at the approved doses vary across different cancers and drug targets (0.17-1.56 L/day), while tend to be similar within a disease/drug target. Anti-CD20 mAbs for B-cell lymphomas show a higher clearance (~ 1 L/day) than other cancers and targets (e.g., ~ 0.3 L/day for anti-PD-1). CONCLUSIONS: Clearance of mAbs can be a tumor/drug target-agnostic biomarker for potential anti-tumor activity as clearance decreases with increasing ORR. Our findings shed important insights into target clearance values that may lead to desired efficacy for different cancers and drug targets, which can be used to guide dose selection for the future development of mAbs during FIH oncology studies.

Prognostic Significance of Tumor-Infiltrating Lymphocytes Determined Using LinkNet on Colorectal Cancer Pathology Images.

PURPOSE: Tumor-infiltrating lymphocytes (TILs) have a significant prognostic value in cancers. However, very few automated, deep learning-based TIL scoring algorithms have been developed for colorectal cancer (CRC). MATERIALS AND METHODS: We developed an automated, multiscale LinkNet workflow for quantifying TILs at the cellular level in CRC tumors using H&E-stained images from the Lizard data set with annotations of lymphocytes. The predictive performance of the automatic TIL scores (TILsLink) for disease progression and overall survival (OS) was evaluated using two international data sets, including 554 patients with CRC from The Cancer Genome Atlas (TCGA) and 1,130 patients with CRC from Molecular and Cellular Oncology (MCO). RESULTS: The LinkNet model provided outstanding precision (0.9508), recall (0.9185), and overall F1 score (0.9347). Clear continuous TIL-hazard relationships were observed between TILsLink and the risk of disease progression or death in both TCGA and MCO cohorts. Both univariate and multivariate Cox regression analyses for the TCGA data demonstrated that patients with high TIL abundance had a significant (approximately 75%) reduction in risk for disease progression. In both the MCO and TCGA cohorts, the TIL-high group was significantly associated with improved OS in univariate analysis (30% and 54% reduction in risk, respectively). The favorable effects of high TIL levels were consistently observed in different subgroups (classified according to known risk factors). CONCLUSION: The proposed deep-learning workflow for automatic TIL quantification on the basis of LinkNet can be a useful tool for CRC. TILsLink is likely an independent risk factor for disease progression and carries predictive information of disease progression beyond the current clinical risk factors and biomarkers. The prognostic significance of TILsLink for OS is also evident.

Predicting microsatellite instability and key biomarkers in colorectal cancer from H&E-stained images: achieving state-of-the-art predictive performance with fewer data using Swin Transformer.

Many artificial intelligence models have been developed to predict clinically relevant biomarkers for colorectal cancer (CRC), including microsatellite instability (MSI). However, existing deep learning networks require large training datasets, which are often hard to obtain. In this study, based on the latest Hierarchical Vision Transformer using Shifted Windows (Swin Transformer [Swin-T]), we developed an efficient workflow to predict biomarkers in CRC (MSI, hypermutation, chromosomal instability, CpG island methylator phenotype, and BRAF and TP53 mutation) that required relatively small datasets. Our Swin-T workflow substantially achieved the state-of-the-art (SOTA) predictive performance in an intra-study cross-validation experiment on the Cancer Genome Atlas colon and rectal cancer dataset (TCGA-CRC-DX). It also demonstrated excellent generalizability in cross-study external validation and delivered a SOTA area under the receiver operating characteristic curve (AUROC) of 0.90 for MSI, using the Molecular and Cellular Oncology dataset for training (N = 1,065) and the TCGA-CRC-DX (N = 462) for testing. A similar performance (AUROC = 0.91) was reported in a recent study, using ~8,000 training samples (ResNet18) on the same testing dataset. Swin-T was extremely efficient when using small training datasets and exhibited robust predictive performance with 200-500 training samples. Our findings indicate that Swin-T could be 5-10 times more efficient than existing algorithms for MSI prediction based on ResNet18 and ShuffleNet. Furthermore, the Swin-T models demonstrated their capability in accurately predicting MSI and BRAF mutation status, which could exclude and therefore reduce samples before subsequent standard testing in a cascading diagnostic workflow, in turn reducing turnaround time and costs.

A robust and lightweight deep attention multiple instance learning algorithm for predicting genetic alterations.

Self-attention mechanism-based algorithms are attractive in digital pathology due to their interpretability, but suffer from computation complexity. This paper presents a novel, lightweight Attention-based Multiple Instance Mutation Learning (AMIML) model to allow small-scale attention operations for predicting gene mutations. Compared to the standard self-attention model, AMIML reduces the number of model parameters by approximately 70%. Using data for 24 clinically relevant genes from four cancer cohorts in TCGA studies (UCEC, BRCA, GBM, and KIRC), we compare AMIML with a standard self-attention model, five other deep learning models, and four traditional machine learning models. The results show that AMIML has excellent robustness and outperforms all the baseline algorithms in the vast majority of the tested genes. Conversely, the performance of the reference deep learning and machine learning models vary across different genes, and produce suboptimal prediction for certain genes. Furthermore, with the flexible and interpretable attention-based pooling mechanism, AMIML can further zero in and detect predictive image patches.

Genome-wide association analyses of leaf rust resistance in cultivated emmer wheat.

KEY MESSAGE: Fifteen and eleven loci, with most loci being novel, were identified to associate with seedling and adult resistances, respectively, to the durum-specific races of leaf rust pathogen in cultivated emmer. Leaf rust, caused by Puccinia triticina (Pt), constantly threatens durum (Triticum turgidum ssp. durum) and bread wheat (Triticum aestivum) production worldwide. A Pt race BBBQD detected in California in 2009 poses a potential threat to durum production in North America because resistance source to this race is rare in durum germplasm. To find new resistance sources, we assessed a panel of 180 cultivated emmer wheat (Triticum turgidum ssp. dicoccum) accessions for seedling resistance to BBBQD and for adult resistance to a mixture of durum-specific races BBBQJ, CCMSS, and MCDSS in the field, and genotyped the panel using genotype-by-sequencing (GBS) and the 9 K SNP (Single Nucleotide Polymorphism) Infinium array. The results showed 24 and nine accessions consistently exhibited seedling and adult resistance, respectively, with two accessions providing resistance at both stages. We performed genome-wide association studies using 46,383 GBS and 4,331 9 K SNP markers and identified 15 quantitative trait loci (QTL) for seedling resistance located mostly on chromosomes 2B and 6B, and 11 QTL for adult resistance on 2B, 3B and 6A. Of these QTL, one might be associated with leaf rust resistance (Lr) gene Lr53, and two with the QTL previously reported in durum or hexaploid wheat. The remaining QTL are potentially associated with new Lr genes. Further linkage analysis and gene cloning are necessary to identify the causal genes underlying these QTL. The emmer accessions with high levels of resistance will be useful for developing mapping populations and adapted durum germplasm and varieties with resistance to the durum-specific races.