Unchained Labs (United States)

Background The 10x Genomics Visium and Xenium platform and assays are now providing sub cellular resolution and mapping transcripts to the tissue morphology to study complex biology from whole transcriptomics to precisely targeted gene panels.These assays are of great significance to clinical research in precision medicine and biomarker discovery.However, third party validations on reproducibility, orthogonal concordance, specificity and sensitivity raise questions that are critical to their use.At BioChain Institute Inc. (BioChain), we are running a series of tests on Xenium v1 as well as Visium Cytassist v2 chemistry, using multiple human carcinoma FFPE tissues to address the specificity and sensitivity questions of these platforms.Methods FFPE tissues from BioChain's repository were screened for quality assessment by our pathologist and molecular scientists for tumor content above 30% as well as the RNA quality.Human carcinoma tissues from Breast, Colon, and Lung with high DV200 scores were selected.An array was constructed with these tissues to fit the Xenium slide's imageable area.Two serial sections were used for the Xenium v1 run per 10x Genomics' protocol.The Off-the shelf Human Multi-Tissue and Cancer Panel was used for targeting 377 genes.After the completion of the Xenium assay, these slides were used for H&E staining and imaging followed by Visium v2 assay on the same section.The H&E images were annotated by the pathologist and integrated with Xenium images using Xenium Explorer.The data was sub-sampled for the different regions per pathologist's annotations of stroma, tumor and immune cell regions.Data from these regions were compared between the serial sections processed using Xenium and Visium platforms to evaluate specificity of selected markers and per gene sensitivity.Sensitivity of Xenium data versus the Visium data was also compared for these specific regions. ResultsWe have previously demonstrated reproducibility of Xenium data between serial sections prepared by different operators as well as concordance between the two assays performed on the same section.In this study, we focus on assessing specificity of gene expression and per gene detection sensitivity through the integration of pathologist annotations of the morphology and spatial gene expression data across multiple types of carcinomas.Conclusions Our findings show the sensitivity and specificity of Xenium and Visium assays to delve further into the heterogeneity of tumor microenvironment and patient stratification for precision medicine and biomarker discovery studies.

We show that the DAG-based consensus protocol Tusk (Danezis, Kokoris-Kogias, Sonnino, and Spiegelman [EuroSys ’22]) does not achieve liveness, under reasonable assumptions which are consistent with its specification. In addition, we give a simple 2-round variation of Tusk with lower latency and strong liveness properties, but with suboptimal resilience. We also show that another 2-round protocol, GradedDAG (Dai, Zhang, Xiao, Yue, Xie, and Jin [Cryptology ePrint Archive, 2024]), also has liveness problems analogous to Tusk.

Few-shot Continual Relation Extraction is a crucial challenge for enabling AI systems to identify and adapt to evolving relationships in dynamic real-world domains. Traditional memory-based approaches often overfit to limited samples, failing to reinforce old knowledge, with the scarcity of data in few-shot scenarios further exacerbating these issues by hindering effective data augmentation in the latent space. In this paper, we propose a novel retrieval-based solution, starting with a large language model to generate descriptions for each relation. From these descriptions, we introduce a bi-encoder retrieval training paradigm to enrich both sample and class representation learning. Leveraging these enhanced representations, we design a retrieval-based prediction method where each sample "retrieves" the best fitting relation via a reciprocal rank fusion score that integrates both relation description vectors and class prototypes. Extensive experiments on multiple datasets demonstrate that our method significantly advances the state-of-the-art by maintaining robust performance across sequential tasks, effectively addressing catastrophic forgetting.

ChatGPT is an artificial intelligence (AI) chatbot that provides users with detailed responses and accurate answers to any questions. It has garnered significant attention after its launch in November 2022. We analyze the returns of AI-themed crypto assets around the launch and widespread attention towards ChatGPT. We reveal significant abnormal returns for AI tokens after the launch of ChatGPT, up to 41% over the course of two weeks. Moreover, 90% of tokens exhibit positive abnormal returns. This suggests that the attention towards ChatGPT and AI in general has transitioned to cryptocurrency markets, resulting in positive price effects for AI-related cryptocurrencies.