Positive Samples Research Articles

A Positive-Unlabeled Metric Learning Framework for Document-Level Relation Extraction with Incomplete Labeling

The goal of document-level relation extraction (RE) is to identify relations between entities that span multiple sentences. Recently, incomplete labeling in document-level RE has received increasing attention, and some studies have used methods such as positive-unlabeled learning to tackle this issue, but there is still a lot of room for improvement. Motivated by this, we propose a positive-augmentation and positive-mixup positive-unlabeled metric learning framework (P3M). Specifically, we formulate document-level RE as a metric learning problem. We aim to pull the distance closer between entity pair embedding and their corresponding relation embedding, while pushing it farther away from the none-class relation embedding. Additionally, we adapt the positive-unlabeled learning to this loss objective. In order to improve the generalizability of the model, we use dropout to augment positive samples and propose a positive-none-class mixup method. Extensive experiments show that P3M improves the F1 score by approximately 4-10 points in document-level RE with incomplete labeling, and achieves state-of-the-art results in fully labeled scenarios. Furthermore, P3M has also demonstrated robustness to prior estimation bias in incomplete labeled scenarios.

Decoupled Contrastive Learning for Long-Tailed Recognition

Supervised Contrastive Loss (SCL) is popular in visual representation learning. Given an anchor image, SCL pulls two types of positive samples, i.e., its augmentation and other images from the same class together, while pushes negative images apart to optimize the learned embedding. In the scenario of long-tailed recognition, where the number of samples in each class is imbalanced, treating two types of positive samples equally leads to the biased optimization for intra-category distance. In addition, similarity relationship among negative samples, that are ignored by SCL, also presents meaningful semantic cues. To improve the performance on long-tailed recognition, this paper addresses those two issues of SCL by decoupling the training objective. Specifically, it decouples two types of positives in SCL and optimizes their relations toward different objectives to alleviate the influence of the imbalanced dataset. We further propose a patch-based self distillation to transfer knowledge from head to tail classes to relieve the under-representation of tail classes. It uses patch-based features to mine shared visual patterns among different instances and leverages a self distillation procedure to transfer such knowledge. Experiments on different long-tailed classification benchmarks demonstrate the superiority of our method. For instance, it achieves the 57.7% top-1 accuracy on the ImageNet-LT dataset. Combined with the ensemble-based method, the performance can be further boosted to 59.7%, which substantially outperforms many recent works. Our code will be released.

Molecular characterization of Cryptosporidium spp. and Giardia duodenalis in pet cats in Henan Province, central China

Cryptosporidium spp. and G. duodenalis often infect humans, cats, and other mammals, causing diarrhea and being responsible for numerous outbreaks of waterborne and foodborne infections worldwide. The rapid increase in the number of pet cats poses a substantial public health risk. However, there were few reports about the infection of Cryptosporidium spp. and G. duodenalis infections in pet cats in Henan Province, central China. Thus, to understand the prevalence and genetic distribution of Cryptosporidium spp. and G. duodenalis in pet cats, and to evaluate the zoonotic potential, possible transmission routes and public health implications of isolates, fecal samples (n = 898) were randomly collected from pet cats in 11 cities in Henan Province, central China. Nested PCR based on the SSU rRNA gene and bg gene was used to the prevalence of Cryptosporidium spp. and G. duodenalis, respectively. The prevalence was 0.8 % (7/898) and 2.0 % (18/898) for Cryptosporidium spp. and G. duodenalis respectively. Additionally, the Cryptosporidium spp. positive isolates were identified as C. parvum subtype IIdA19G1 by gp60 gene. In the present study, the IIdA19G1 subtype was discovered in pet cats for the first time in China, enriching the information on the host type and geographical distribution of Cryptosporidium spp. in China. For G. duodenalis, a total of 18 G. duodenalis positive samples were identified, belonging to four assemblages: a zoonotic assemblage A1 (4/898), three host-specific assemblages C (8/898), D (5/898), and F (1/898). Interestingly, we found that pet cats infected with Cryptosporidium spp. and G. duodenalis are more likely to experience emaciation symptoms compared to the negative group. More importantly, the prevalence of Cryptosporidium spp. and G. duodenalis detected in the present study were low, but the subtype IIdA19G1 of Cryptosporidium spp. and the assemblages A1, C, D, and F of G. duodenalis have the potential for zoonotic transmission. Thus, we should focus on preventing and controlling the risk of cross-species transmission that may occur in pet cats in Henan Province.

Plug-In Diffusion Model for Sequential Recommendation

Pioneering efforts have verified the effectiveness of the diffusion models in exploring the informative uncertainty for recommendation. Considering the difference between recommendation and image synthesis tasks, existing methods have undertaken tailored refinements to the diffusion and reverse process. However, these approaches typically use the highest-score item in corpus for user interest prediction, leading to the ignorance of the user's generalized preference contained within other items, thereby remaining constrained by the data sparsity issue. To address this issue, this paper presents a novel Plug-in Diffusion Model for Recommendation (PDRec) framework, which employs the diffusion model as a flexible plugin to jointly take full advantage of the diffusion-generating user preferences on all items. Specifically, PDRec first infers the users' dynamic preferences on all items via a time-interval diffusion model and proposes a Historical Behavior Reweighting (HBR) mechanism to identify the high-quality behaviors and suppress noisy behaviors. In addition to the observed items, PDRec proposes a Diffusion-based Positive Augmentation (DPA) strategy to leverage the top-ranked unobserved items as the potential positive samples, bringing in informative and diverse soft signals to alleviate data sparsity. To alleviate the false negative sampling issue, PDRec employs Noise-free Negative Sampling (NNS) to select stable negative samples for ensuring effective model optimization. Extensive experiments and analyses on four datasets have verified the superiority of the proposed PDRec over the state-of-the-art baselines and showcased the universality of PDRec as a flexible plugin for commonly-used sequential encoders in different recommendation scenarios. The code is available in https://github.com/hulkima/PDRec.

Review-Enhanced Hierarchical Contrastive Learning for Recommendation

Designed to establish potential relations and distill high-order representations, graph-based recommendation systems continue to reveal promising results by jointly modeling ratings and reviews. However, existing studies capture simple review relations, failing to (1) completely explore hidden connections between users (or items), (2) filter out redundant information derived from reviews, and (3) model the behavioral association between rating and review interactions. To address these challenges, we propose a review-enhanced hierarchical contrastive learning, namely ReHCL. First, ReHCL constructs topic and semantic graphs to fully mine review relations from different views. Moreover, a cross-view graph contrastive learning is used to achieve enhancement of node representations and extract useful review knowledge. Meanwhile, we design a neighbor-based positive sampling to capture the graph-structured similarity between topic and semantic views, further performing efficient contrast and reducing redundant noise. Next, we propose a cross-modal contrastive learning to match the rating and review representations, by exploring the association between ratings and reviews. Lastly, these two contrastive learning modes form a hierarchical contrastive learning task, which is applied to enhance the final recommendation task. Extensive experiments verify the superiority of ReHCL compared with state-of-the-arts.

Multi-Label Supervised Contrastive Learning

Multi-label classification is an arduous problem given the complication in label correlation. Whilst sharing a common goal with contrastive learning in utilizing correlations for representation learning, how to better leverage label information remains challenging. Previous endeavors include extracting label-level presentations or mapping labels to an embedding space, overlooking the correlation between multiple labels. It exhibits a great ambiguity in determining positive samples with different extent of label overlap between samples and integrating such relations in loss functions. In our work, we propose Multi-Label Supervised Contrastive learning (MulSupCon) with a novel contrastive loss function to adjust weights based on how much overlap one sample shares with the anchor. By analyzing gradients, we explain why our method performs better under multi-label circumstances. To evaluate, we conduct direct classification and transfer learning on several multi-label datasets, including widely-used image datasets such as MS-COCO and NUS-WIDE. Validation indicates that our method outperforms the traditional multi-label classification method and shows a competitive performance when comparing to other existing approaches.

KGTS: Contrastive Trajectory Similarity Learning over Prompt Knowledge Graph Embedding

Trajectory similarity computation serves as a fundamental functionality of various spatial information applications. Although existing deep learning similarity computation methods offer better efficiency and accuracy than non-learning solutions, they are still immature in trajectory embedding and suffer from poor generality and heavy preprocessing for training. Targeting these limitations, we propose a novel framework named KGTS based on knowledge graph grid embedding, prompt trajectory embedding, and unsupervised contrastive learning for improved trajectory similarity computation. Specifically, we first embed map grids with a GRot embedding method to vigorously grasp the neighbouring relations of grids. Then, a prompt trajectory embedding network incorporates the resulting grid embedding and extracts trajectory structure and point order information. It is trained by unsupervised contrastive learning, which not only alleviates the heavy preprocessing burden but also provides exceptional generality with creatively designed strategies for positive sample generation. The prompt trajectory embedding adopts a customized prompt paradigm to mitigate the gap between the grid embedding and the trajectory embedding. Extensive experiments on two real-world trajectory datasets demonstrate the superior performance of KGTS over state-of-the-art methods.

AQ-DETR: Low-Bit Quantized Detection Transformer with Auxiliary Queries

DEtection TRansformer (DETR)-based models have achieved remarkable performance. However, they are accompanied by a large computation overhead cost, which significantly prevents their applications on resource-limited devices. Prior arts attempt to reduce the computational burden of DETR using low-bit quantization, while these methods sacrifice a severe significant performance on weight-activation-attention low-bit quantization. We observe that the number of matching queries and positive samples affect much on the representation capacity of queries in DETR, while quantifying queries of DETR further reduces its representational capacity, thus leading to a severe performance drop. We introduce a new quantization strategy based on Auxiliary Queries for DETR (AQ-DETR), aiming to enhance the capacity of quantized queries. In addition, a layer-by-layer distillation is proposed to reduce the quantization error between quantized attention and full-precision counterpart. Through our extensive experiments on large-scale open datasets, the performance of the 4-bit quantization of DETR and Deformable DETR models is comparable to full-precision counterparts.

SALSA: Semantically-Aware Latent Space Autoencoder

In deep learning for drug discovery, molecular representations are often based on sequences, known as SMILES, which allow for straightforward implementation of natural language processing methodologies, one being the sequence-to-sequence autoencoder. However, we observe that training an autoencoder solely on SMILES is insufficient to learn molecular representations that are semantically meaningful, where semantics are specified by the structural (graph-to-graph) similarities between molecules. We demonstrate by example that SMILES-based autoencoders may map structurally similar molecules to distant codes, resulting in an incoherent latent space that does not necessarily respect the semantic similarities between molecules. To address this shortcoming we propose Semantically-Aware Latent Space Autoencoder (SALSA) for molecular representations: a SMILES-based transformer autoencoder modified with a contrastive task aimed at learning graph-to-graph similarities between molecules. To accomplish this, we develop a novel dataset comprised of sets of structurally similar molecules and opt for a supervised contrastive loss that is able to incorporate full sets of positive samples. We evaluate semantic awareness of SALSA representations by comparing to its ablated counterparts, and show empirically that SALSA learns representations that maintain 1) structural awareness, 2) physicochemical awareness, 3) biological awareness, and 4) semantic continuity.

Debiasing Multimodal Sarcasm Detection with Contrastive Learning

Despite commendable achievements made by existing work, prevailing multimodal sarcasm detection studies rely more on textual content over visual information. It unavoidably induces spurious correlations between textual words and labels, thereby significantly hindering the models' generalization capability. To address this problem, we define the task of out-of-distribution (OOD) multimodal sarcasm detection, which aims to evaluate models' generalizability when the word distribution is different in training and testing settings. Moreover, we propose a novel debiasing multimodal sarcasm detection framework with contrastive learning, which aims to mitigate the harmful effect of biased textual factors for robust OOD generalization. In particular, we first design counterfactual data augmentation to construct the positive samples with dissimilar word biases and negative samples with similar word biases. Subsequently, we devise an adapted debiasing contrastive learning mechanism to empower the model to learn robust task-relevant features and alleviate the adverse effect of biased words. Extensive experiments show the superiority of the proposed framework.

Diversity of Trichinella species in carnivores from Bosnia and Herzegovina

BackgroundIn Bosnia and Herzegovina, domestic and wild carnivores represent a significant driver for the transmission and ecology of zoonotic pathogens, especially those of parasitic aetiology. Nevertheless, there is no systematic research of Trichinella species in animals that have been conducted in Bosnia and Herzegovina, even though trichinellosis is considered the most important parasitic zoonosis. The available results of the few studies carried out in Bosnia and Herzegovina are mainly related to the confirmation of parasitic larvae in the musculature of domestic pigs and wild boars or data related to trichinellosis in humans. The objective of our study was to present the findings of a comprehensive investigation into the species composition of Trichinella among 11 carnivorous species within the territory of Bosnia and Herzegovina, as follows: red fox (Vulpes vulpes), grey wolf (Canis lupus), brown bear (Ursus arctos), wildcat (Felis silvestris), pine marten (Martes martes), European badger (Meles meles), weasel (Mustela nivalis), European polecat (Mustela putorius), Eurasian lynx (Lynx lynx), but also dog (Canis lupus familiaris) and cat (Felis catus).ResultsIn the period 2013–2023, carnivore musculature samples (n = 629), each consisting of 10 g of muscle tissue, were taken post-mortem and individually examined using the artificial digestion method. In the positive samples (n = 128), molecular genotyping and identification of parasitic larvae of Trichinella spp. were performed using a PCR-based technique up to the species/genotype level. Positive samples were used for basic PCR detection of the genus Trichinella (rrnS rt-PCR technique) and genotyping (rrnl-EVS rt-PCR technique). The Trichinella infection was documented for the first time in Bosnia and Herzegovina among red foxes, grey wolves, brown bears, dogs, badgers and Eurasian lynx, with a frequency rate of 20.3%. Additionally, the presence of T. britovi infection was newly confirmed in Bosnia and Herzegovina, marking the initial documented cases. Furthermore, both T. britovi and T. pseudospiralis infections were observed in the wildcat population, whereas T. britovi and T. spiralis infections were detected in pine martens. Consistent with previous research, our findings align particularly regarding carnivores, with data from other countries such as Germany, Finland, Romania, Poland and Spain, where T. britovi exhibits a wider distribution (62.5–100%) compared to T. spiralis (0.0–37.5%). T. britovi is more common among sylvatic carnivores (89.0%), while T. spiralis prevails in wild boars (62.0%), domestic swine (82.0%) and rodents (75.0%).ConclusionThe results of our study represent the first molecular identification of species of the genus Trichinella in Bosnia and Herzegovina. Additionally, our findings underscore the necessity for targeted epidemiological studies to thoroughly assess trichinellosis prevalence across diverse animal populations. Considering the relatively high frequency of trichinellosis infection in investigated animal species and its public health implications, there is an evident need for establishing an effective trichinellosis surveillance system in Bosnia and Herzegovina.

Screening for Selenomonas noxia in a Pediatric and Adolescent Patient Population Reveals Differential Oral Prevalence across Age Groups.

Selenomonas noxia, a gram-negative anaerobe usually present in periodontitis, may be linked to overweight and obese adults. Recent advancements include a valid qPCR screening, enabling an effective prevalence study among pediatric patients aged 7 to 17 years. The aim of this study was to complete a retrospective screening of saliva samples from an existing biorepository using a validated qPCR screening protocol. The pediatric study sample (n = 87) comprised nearly equal numbers of males and females, mostly minority patients (67%), with an average age of 13.2 years. Screening for Selenomonas noxia revealed 34.4% (n = 30/87) positive samples, evenly distributed between males and females (p = 0.5478). However, an age-dependent association was observed with higher percentages of positive samples observed with higher ages (13.3% among 7 to 10 years; 34.6% among 11 to 13 years; 54.8% among 14-17 years), which was statistically significant (p = 0.0001). Although these findings revealed no noteworthy distinctions between males or females and minorities and non-minorities, the notable contrast between younger (7 to 10 years) and older (11 to 17 years) participants, possibly influenced by factors such as hormones and behavioral traits, will require further investigation of this patient population.

Occurrence of ochratoxin A in Sichuan bacon from different geographical regions and characterization and biocontrol of ochratoxigenic Aspergillus westerdijkiae strain 21G2-1A

Sichuan bacon represents the most prevalent dry-cured meat product across Southwest China, but it is vulnerable to fungal spoilage. In the present study, a total of 47 Sichuan bacons were obtained from different regions of the Sichuan Province and analyzed for the presence of ochratoxin A (OTA), yielding a positive rate of 23.4 % (11/47). All the observed OTA concentrations exceeded the maximum admissible dose in meat products (1 μg/kg) established by some EU countries, with the highest OTA concentration being 250.75 μg/kg, which raises a food safety concern and reveals the need for a standardized scientific processing protocol. Then, an OTA-producing fungus named 21G2-1A was isolated from positive samples and found to be Aspergillus westerdijkiae. Further characterization suggested a positive correlation between fungal growth and OTA production. The optimal temperature for the former was 25 °C, while it was 20 °C for the latter. Although the A. westerdijkiae strain 21G2-1A demonstrated greater mycelium growth in the presence of NaCl, OTA production was significantly dismissed when the salinity was greater than 5 %. Four lactic acid bacteria (LAB) were screened out as antagonists against the ochratoxigenic fungus. In vitro evaluation of the antagonists revealed that live cells inhibited fungal growth, and adsorption also contributed to OTA removal at different levels. This study sheds some light on OTA control in Sichuan bacon through a biological approach.

Abstract 4768: Differential expression of triptolide gene targets in human papillomavirus positive and negative head and neck squamous cell carcinomas

Abstract Head and neck squamous cell carcinoma (HNSCC) is the fifth most common cancer in the US. HPV represents a more recent risk factor for which treatment may be individualized. HPV expresses viral E6 and E7 oncoproteins which downregulate wild type p53 and Rb respectively. Triptolide, a derivative of a traditional Chinese medicinal herb, has been shown to inhibit viral E6, thereby reactivating wild type p53. Therefore, triptolide may target HPV positive HNSCC as a putative treatment. To explore this, we compared mutation, methylation, and expression of triptolide gene targets between HPV positive and HPV negative samples from The Cancer Genome Atlas (TCGA) program. Patient data was queried from the TCGA PanCancer Atlas for HNSCC using cBioPortal. This included a total of 523 patients with 487 meeting inclusion criteria. 415 patients were HPV (-) and 72 were HPV (+).Differences in direct genomic alterations, DNA methylation, mRNA expression, and protein expression of the 34 known triptolide targets were compared between these two groups using the Benjamini-Hochberg Procedure with a significance level of q < 0.05. Significant differences of the studied genes (q < 0.05) are shown in Table 1. There were no significant differences in protein expression. Of the 34 known triptolide targets, HPV positive and negative samples show variable enrichment. Triptolide targets exhibit greater methylation and mRNA expression in HPV positive samples supporting triptolide’s efficacy in these samples. However, other triptolide targets are the differentially expressed and methylated in HPV negative samples, so the compound may have a beneficial effect in this context as well. Further studies are required to test efficacy of triptolide against HPV positive and HPV negative HNSCC in vitro and in vivo models. Table 1. Significant genomic differences between HPV (+) and HPV (-) samples shown. Enriched in HPV (+) Enriched in HPV (-) Gene Log2 ratio q-value Gene Log2 ratio q-value DNA methylation Direct genomic mutations VEGFA 0 5.85E-03 TP53 3.08 4.53E-29 BCL2 0.07 9.78E-18 DNA methylation JUN 0 4.89E-02 RELA 0.07 5.06E-04 CD80 0.06 2.56E-02 STAT3 0.24 3.19E-20 BIRC3 0.01 1.29E-04 TNF 0.07 4.31E-04 CD274 0.07 4.40E-03 XIAP 0.07 8.02E-06 CCR7 0.11 3.56E-08 CXCR4 0.04 3.58E-03 CD1A 0.10 2.36E-05 CD40 0.06 1.00E-05 C3 0.06 6.91E-03 CD14 0.04 8.12E-03 CD86 0.03 2.45E-02 mRNA expression mRNA expression RELA 0.16 1.90E-03 STAT3 0.25 4.22E-03 CDKN1A 0.28 3.00E-02 BCL2 1.95 1.68E-14 PLAU 1.37 3.24E-12 CASP3 0.45 2.55E-07 CXCL8 1.05 1.51E-04 TP53 1.45 3.16E-25 MCL1 0.23 3.59E-04 IL2 0.52 2.13E-06 TGFB1 0.77 3.27E-11 INFG 1.15 2.08E-05 DEFB4A 1.10 1.77E-02 CXCR4 1.18 8.03E-07 BIRC3 1.20 6.14E-06 IL23A 0.77 3.88E-04 CCR7 1.18 2.10E-06 CD40 0.54 2.27E-04 Citation Format: William E. Kamm, Lindsey J. Mortenson, Frank G. Ondrey. Differential expression of triptolide gene targets in human papillomavirus positive and negative head and neck squamous cell carcinomas [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4768.

Overlapping object detection with adaptive Gaussian sample division and asymmetric weighted loss

Existing deep learning based detectors are mostly designed for scenes with sparsely distributed objects. However, in certain scenarios such as dense crowds, objects often overlap severely. The dense anchor arrangement in anchor-based detectors is not quite suitable for the overlapping object detection. Anchor-free detectors have the potential to achieve high-performance in overlapping object detection, but troubled by the extreme imbalance of positive and negative samples. To this end, we propose an anchor-free overlapping object detector. Our adaptive Gaussian sample division (AGSD) can effectively allocate positive and negative samples with clear semantics to overlapping objects. Secondly, asymmetric weighted loss (AW Loss) adapts to continuous positive and negative sample values, thereby improving the classification ability of the detector. Lastly, our global location distribution head (GLD head) can introduce the supervision of overlapping object distributions. To verify the effectiveness of our method, we construct a large-scale high-quality overlapping object detection dataset containing 6173 images and 17,725 annotations. Compared with mainstream object detector, our method achieves the best performance of AP50 at 96.71%.

Prevalence of Cryptosporidium spp. in Sheep and Goats in Jiangsu, China.

Sheep and goats serve as crucial hosts for Cryptosporidium spp. and are primarily responsible for its transmission via the fecal-oral route. This can result in symptoms such as lamb weight loss, diarrhea, and even fatalities, leading to significant economic losses. Currently, there is a lack of scholarly research investigating the prevalence of Cryptosporidium spp. infection in sheep and goats specifically within Jiangsu province. This study collected fecal samples from sheep and goats, extracted their DNA, amplified target bands using nested PCR, sequenced the DNA, constructed a phylogenetic tree, and identified the genetic genotype. In total, 3 positive samples were identified out of 398 samples. Furthermore, the gene sequences of these samples exhibited significant homology with C. xiaoi in GenBank. The phylogenetic analysis revealed that the Cryptosporidium spp. parasites under investigation are phylogenetically related to C. xiaoi. Conducting epidemiological investigations and accurately identifying the species of Cryptosporidium spp. is of utmost importance not only for the mutton sheep farming industry in Jiangsu but also for the proactive safeguarding of human health.

Abstract 5023: Analytical validation of the Oncomine™ Dx Express Test using liquid biopsy samples as an IUO assay in a CAP/CLIA lab

Abstract Introduction: The Oncomine™ Dx Express Test (ODxET) is a next-generation sequencing (NGS) test. It can be used as an investigational use only (IUO**) test to detect somatic DNA and RNA alterations in human cell-free total nucleic acid (cfTNA) isolated from plasma samples. This study outlines the analytical validation of the ODxET for relevant single nucleotide variants (SNV) and insertion/deletions (INDEL) in liquid biopsy specimens. The Ion Torrent Genexus™ platform has a short turnaround time with minimal operator input. This study was performed in a College of American Pathologists (CAP™) accredited and Clinical Laboratory Improvement Act (CLIA) licensed laboratory. Methods: Forty-seven unique samples (39 clinical, 3 derived, 2 reference and 3 commercially available DNA controls) were analyzed. All experiments began with sample extraction, except the lower input limit of detection (LLOD). We evaluated limit of detection, variant calling accuracy, and precision. Two control NIST™ samples were sequenced with 3 technical replicates each for INDEL and SNVs analytical accuracy. The LOD was evaluated at both the target input of 2ng/µl (40ng), and the minimum input,0.33ng/µl (6.6ng) with allele frequency (AF) levels at 1%, 0.5%, and 0.25%. The 39 clinical plasma samples were used for variant calling accuracy. Inter-assay reproducibility was assessed by sequencing DNA controls across four runs by different operators over multiple days/systems. Intra-assay repeatability was assessed by sequencing 4 samples in replicate on the same run by one operator on the same day/system. Results: The analytical accuracy and NPV was 100% for SNV/INDEL variants. The 40ng input LOD sensitivity was 96% for SNVs at 0.5% AF and 100% for INDELs at 0.25% Minimum AF. The lower input LOD sensitivity was 89% for SNVs and 100% for INDEL at 1% AF. Variant calling accuracy on samples with pre-characterized data was 100% for the 13 positive samples. Specificity was >99% for all samples including both healthy donor and positive samples. Repeatability was 100% for INDELs and >96% for SNVs, reproducibility was 100% for both variant classes. Conclusion: This validation study to estimate analytical accuracy, limit of detection, variant calling accuracy and precision has demonstrated that the ODxET assay is highly robust in detecting SNV and INDEL mutations in plasma samples and is suitable for use in a CAP/CLIA laboratory as an IUO assay.**=In Vitro Diagnostic use only. Not available in all regions including North America. Citation Format: Juan C. Espinoza, Rajendra Ramsamooj, Akemi Yuki, John Williamson, Diarra Hassell, Madhu Jasti, Suzanne Salazar, David Ginzinger. Analytical validation of the Oncomine™ Dx Express Test using liquid biopsy samples as an IUO assay in a CAP/CLIA lab [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 5023.

Abstract 1399: The role of CCN5 in regulation of tumor-infiltrating lymphocytes (TILS) in triple-negative breast cancer (TNBC) cells

Abstract TNBC accounts for about 15-20 percent of all breast cancer cases, with a high incidence in young and African-American women. TNBC lacks the expression of estrogen receptor, progesterone receptor, and normal human epidermal growth factor receptor type 2 (HER2) copy number with expression. Thus, TNBC is not sensitive to endocrine and targeted molecular therapies. Therefore, surgery and systemic chemotherapy are the primary treatment methods for TNBC. The adjuvant therapy regimen includes anthracycline, paclitaxel, or both in (Neo-) early TNBC. However, this treatment is often ineffective after recurrence or metastasis. Therefore, continuously searching for new targeted therapies and immunotherapies becomes a focus in TNBC treatment in patients with diverse communities. The studies showed that TNBC is more immunogenic than other subtypes of breast cancers with tumor-infiltrating lymphocytes (TILs) in its microenvironment. More importantly, Cytotoxic CD8+ TILs of the adaptive immune system are the most potent effectors in the anticancer immune response and constitute the backbone of cancer immunotherapy. Therefore, a high number of CD8+TILs in TNBC patients' tumors are considered a better prognosis following neoadjuvant chemotherapy. Hence, increasing and activating CD8+ TILs could be an exciting and appropriate approach to treating TNBC. Blocking immune checkpoints can inverse the microenvironment immunosuppressive state and enhance the function of tumor cell clearance. One of the major immune checkpoint therapies has involved a monoclonal antibody against programmed cell death-1 (PD-1)/programmed cell death-ligand 1 (PD-L1), expressed in various tumor cells, including TNBC and immune cells. The studies found that PD-1/PD-L1 negatively regulates CD8+ TILs in cancers, and blockade of PD-1 can induce CD8+ TILs responses in advanced melanoma. Interestingly, PD-1/PD-L1 immunotherapy has also progressed in suppressing TNBC growth and progression. However, the currently available immune checkpoint therapies have therapeutic limitations with the toxicity of the drugs. Thus, new therapeutic options are urgently needed. Our studies found that the expression of CCN5, an anti-invasive gene, correlates with the number of cytotoxic CD8+ TILs in ER+ and HER-2 positive breast cancer samples. Furthermore, we also observed that PD-L1 (programmed death-ligand 1), which was found to have high and low expression in MDA-MB-231 and 4T1, respectively, was suppressed in MDA-MB-231 and 4T1 TNBC cell lines by CCN5 recombinant protein treatment for 48 h. These findings might provide a novel immune checkpoint mechanistic explanation of CCN5 that mediates TNBC growth and progression suppression. Citation Format: Upasana Biswas, Anjali Kambhampati, Noor Haideri, Archana De, Sunil Upadhyay, Snigdha Banerjee, Sushanta Banerjee. The role of CCN5 in regulation of tumor-infiltrating lymphocytes (TILS) in triple-negative breast cancer (TNBC) cells [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1399.

Abstract 4629: QClampTM Plex VEXAS syndrome UBA1 mutation detection assay: A noval XNA based assay to simultaneously detect eight UBA1 somatic mutations associated with VEXAS syndrome

Abstract VEXAS syndrome (acronym for Vacuoles, E1 enzyme, X-linked, Autoinflammatory, Somatic) is a monogenic disease discovered in December 2020. It is an adult-onset systemic inflammatory disorder genetically characterized by somatic mutation of the UBA1 gene which encodes the ubiquitin like modifier activating enzyme 1. VEXAS patients have overlapping rheumatologic and hematologic manifestations, including increased risk for myelodysplastic syndrome (MDS). It is critically important to diagnose VEXAS syndrome, but UBA1 mutation test is still not commercially available. The three most frequent mutations of UBA1 gene are p.M41T(122T>C), p.M41V (c.121A>G), and p.M41L (c.121A>C) in codon 41 of exon 3. Other mutations including splice site mutations at exon 3 (c.118-2A>C, c.118-1G>C and c.118- 9_118-2del), p.S56F (c.167C>T) in codon 56 of exon 3, and p.S621C (c.1861A>T) in codon 621 of exon 16 have been reported. Based on XNA technology, we developed a QClampTM Plex platform to detect all the eight mutations in a single reaction with Luminex system. XNA technology enriches mutant target sequence in a PCR reaction by using synthetic DNA analog XNA (xenonucleic acid) to block wild type DNA. With spiking synthetic mutant DNA in GM24385 cell line DNA, we determined that this assay can detect mutations with a detection limit of variant allele frequency (VAF) for M41T at 0.1%, M41V at 0.1%, M41L at 0.25%, c.118-2A>C at 1%, c.118-1G>C at 1%, c.118- 9_118-2del at 1%, S56F at 0.25% and S621C at 1%, respectively, and specificity of 100% for all mutations. We tested saved DNA samples extracted from whole blood previously collected from suspected VEXAS patients and tested at the National Cancer Institute by Sanger sequencing, including 13 positive and 20 negative samples. Our assay detected all corresponding UBA1 mutations in the 13 positive patient samples, and did not detect UBA1 mutation in 18 of the 20 negative patient samples. Two of the 20 negative samples were also tested positive for UBA1 M41T and M41L on our assay, respectively, due to its significantly higher sensitivity than the Sanger sequencing platform, which has a detection limit of 15-20% VAF. Citation Format: Yunqing Ma. QClampTM Plex VEXAS syndrome UBA1 mutation detection assay: A noval XNA based assay to simultaneously detect eight UBA1 somatic mutations associated with VEXAS syndrome [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4629.

Abstract 6077: Chimeric RNAs are abundantly expressed in precancerous breast tissue of sporadic breast cancer patients

Abstract Background: Immuno-prevention efforts in breast cancer (BC) have been hampered by the lack of known neoantigenic targets that can be used for a cancer prevention vaccine. RNA fusions represent attractive candidates as these are more immunogenic than insertion/deletion mutations and are substantial source of neoantigens. The objective of this study was to characterize chimeric RNAs in precancerous breast tissue. Methodology: Novel RNA fusions were identified by mining the sequence reads from RNA sequencing of 25 Her2 positive, 25 hormone receptor positive (HR) and 25 triple negative (TN) breast tumors and paired histologically normal tissues by Qiagen’s CLC pipeline. Breast samples from women undergoing breast reduction surgery were used as controls to exclude normal tissue associated chimeric mRNAs from analysis. Neoantigen prediction was performed by MHCnuggets. Chimeric mRNAs were validated by PCR. Ribosomal profiling was performed to assess if fusion transcripts are engaged with ribosomes. Results: A median of 36 novel RNA fusions (range 7-877) were identified in the tumor samples. In line with high genomic instability of TNBC subtype of BC, about 1/3 of the TN tumors expressed greater than 300 RNA fusions. In order to identify RNA fusions that are relevant for BC prevention, we investigated the presence of RNA fusions in the paired histologically normal samples as compared to the index tumor. We restricted these analyses to top 20 fusion transcripts that were present across the set of 75 tumor samples and also detected in 1 or more of the TCGA samples. Interestingly, we found that more than 1/3 of the histologically normal samples express at least 1 of the top 20 RNA fusions. When compared to the corresponding tumor, we found up to 50% of the paired normal tissue expressed RNA fusions identified in the tumor, supporting a role for these chimeric mRNA as an early molecular change during breast tumorigenesis. To validate the robustness of the prediction pipeline, PCR validation was performed for the chimeric mRNA with highest prevalence, NSFP1- LRRC37A2 which confirmed the presence of the transcript in 72% (13/18) of the predicted positive patient samples. MHCnuggets pipeline predicted 15 neopeptides from intergenically spliced chimeric NSFP1 [Exon 1-13] - LRRC37A2 [Exon 2-14], ¼ of which we validated to be immunogenic by ELISA. In order to confirm that the translation of the chimeric NSFP1- LRRC37A2 transcript, we investigated their enrichment in polysome enriched fraction obtained from a BC cell line confirmed to express the fusion transcript by PCR. Enrichment of transcripts in the polysomal fraction confirmed that these transcripts are likely to be translated. Conclusions: RNA fusions are frequently present in at risk breast tissue and are the source of substantial number of neoantigens. These RNA fusion- derived neoantigens may provide novel opportunities for developing vaccines for BC prevention. Citation Format: Anjana Bhardwaj, Shiyanth Thevasagayampillai, Sakuni Rankothgedera, Prisha Verma, Micah B. Castillo, Alexander C. Koh, Constance Albarracin, Preethi H. Gunaratne, Isabelle Bedrosian. Chimeric RNAs are abundantly expressed in precancerous breast tissue of sporadic breast cancer patients [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6077.