Search

In the face of uncertainty, the ability to *seek information* is of fundamental importance. In many practical applications, such as medical diagnosis and troubleshooting, the information needed to solve the task is not initially given and has to be actively sought by asking follow-up questions (for example, a doctor asking a patient for more details about their symptoms). In this work, we introduce Uncertainty of Thoughts (UoT), an algorithm to augment large language models with the ability to actively seek information by asking effective questions. UoT combines 1) an *uncertainty-aware simulation approach* which enables the model to simulate possible future scenarios and how likely they are to occur, 2) *uncertainty-based rewards* motivated by information gain which incentivizes the model to seek information, and 3) a *reward propagation scheme* to select the optimal question to ask in a way that maximizes the expected reward. In experiments on medical diagnosis, troubleshooting, and the `20 Questions` game, UoT achieves an average performance improvement of 38.1% in the rate of successful task completion across multiple LLMs compared with direct prompting and also improves efficiency (i.e., the number of questions needed to complete the task). Our code has been released [here](https://github.com/zhiyuanhubj/UoT), Comment: Update Results

Seeking answers effectively for long videos is essential to build video question answering (videoQA) systems. Previous methods adaptively select frames and regions from long videos to save computations. However, this fails to reason over the whole sequence of video, leading to sub-optimal performance. To address this problem, we introduce a state space layer (SSL) into multi-modal Transformer to efficiently integrate global semantics of the video, which mitigates the video information loss caused by frame and region selection modules. Our SSL includes a gating unit to enable controllability over the flow of global semantics into visual representations. To further enhance the controllability, we introduce a cross-modal compositional congruence (C^3) objective to encourage global semantics aligned with the question. To rigorously evaluate long-form videoQA capacity, we construct two new benchmarks Ego-QA and MAD-QA featuring videos of considerably long length, i.e. 17.5 minutes and 1.9 hours, respectively. Extensive experiments demonstrate the superiority of our framework on these new as well as existing datasets., Comment: Work in progress

Dialogue systems have received increasing attention while automatically evaluating their performance remains challenging. User satisfaction estimation (USE) has been proposed as an alternative. It assumes that the performance of a dialogue system can be measured by user satisfaction and uses an estimator to simulate users. The effectiveness of USE depends heavily on the estimator. Existing estimators independently predict user satisfaction at each turn and ignore satisfaction dynamics across turns within a dialogue. In order to fully simulate users, it is crucial to take satisfaction dynamics into account. To fill this gap, we propose a new estimator ASAP (sAtisfaction eStimation via HAwkes Process) that treats user satisfaction across turns as an event sequence and employs a Hawkes process to effectively model the dynamics in this sequence. Experimental results on four benchmark dialogue datasets demonstrate that ASAP can substantially outperform state-of-the-art baseline estimators., Comment: To appear at ACL 2023

The problem of continual learning has attracted rising attention in recent years. However, few works have questioned the commonly used learning setup, based on a task curriculum of random class. This differs significantly from human continual learning, which is guided by taxonomic curricula. In this work, we propose the Taxonomic Class Incremental Learning (TCIL) problem. In TCIL, the task sequence is organized based on a taxonomic class tree. We unify existing approaches to CIL and taxonomic learning as parameter inheritance schemes and introduce a new such scheme for the TCIL learning. This enables the incremental transfer of knowledge from ancestor to descendant class of a class taxonomy through parameter inheritance. Experiments on CIFAR-100 and ImageNet-100 show the effectiveness of the proposed TCIL method, which outperforms existing SOTA methods by 2% in terms of final accuracy on CIFAR-100 and 3% on ImageNet-100.

The problem of class incremental learning (CIL) is considered. State-of-the-art approaches use a dynamic architecture based on network expansion (NE), in which a task expert is added per task. While effective from a computational standpoint, these methods lead to models that grow quickly with the number of tasks. A new NE method, dense network expansion (DNE), is proposed to achieve a better trade-off between accuracy and model complexity. This is accomplished by the introduction of dense connections between the intermediate layers of the task expert networks, that enable the transfer of knowledge from old to new tasks via feature sharing and reusing. This sharing is implemented with a cross-task attention mechanism, based on a new task attention block (TAB), that fuses information across tasks. Unlike traditional attention mechanisms, TAB operates at the level of the feature mixing and is decoupled with spatial attentions. This is shown more effective than a joint spatial-and-task attention for CIL. The proposed DNE approach can strictly maintain the feature space of old classes while growing the network and feature scale at a much slower rate than previous methods. In result, it outperforms the previous SOTA methods by a margin of 4\% in terms of accuracy, with similar or even smaller model scale., Comment: Accepted by CVPR2023

Fully fine-tuning pretrained large-scale transformer models has become a popular paradigm for video-language modeling tasks, such as temporal language grounding and video-language summarization. With a growing number of tasks and limited training data, such full fine-tuning approach leads to costly model storage and unstable training. To overcome these shortcomings, we introduce lightweight adapters to the pre-trained model and only update them at fine-tuning time. However, existing adapters fail to capture intrinsic temporal relations among video frames or textual words. Moreover, they neglect the preservation of critical task-related information that flows from the raw video-language input into the adapter's low-dimensional space. To address these issues, we first propose a novel REcurrent ADapter (READ) that employs recurrent computation to enable temporal modeling capability. Second, we propose Partial Video-Language Alignment (PVLA) objective via the use of partial optimal transport to maintain task-related information flowing into our READ modules. We validate our READ-PVLA framework through extensive experiments where READ-PVLA significantly outperforms all existing fine-tuning strategies on multiple low-resource temporal language grounding and video-language summarization benchmarks., Comment: Accepted at AAAI 2024

Large Language Models (LLMs) have marked a significant advancement in the field of natural language processing, demonstrating exceptional capabilities in reasoning, tool usage, and memory. As their applications extend into multi-agent environments, a need has arisen for a comprehensive evaluation framework that captures their abilities in reasoning, planning, collaboration, and more. This work introduces a novel benchmarking framework specifically tailored to assess LLMs within multi-agent settings, providing quantitative metrics to evaluate their judgment, reasoning, deception, self-awareness, cooperation, coordination, and rationality. We utilize games such as Chameleon and Undercover, alongside game theory scenarios like Cost Sharing, Multi-player Prisoner's Dilemma, and Public Good, to create diverse testing environments. Our framework is fortified with the Probabilistic Graphical Modeling (PGM) method, enhancing the LLMs' capabilities in navigating complex social and cognitive dimensions. The benchmark evaluates seven multi-agent systems powered by different LLMs, quantitatively highlighting a significant capability gap over threefold between the strongest, GPT-4, and the weakest, Llama-2-70B. It also confirms that our PGM enhancement boosts the inherent abilities of all selected models by 50% on average. Our codes are released here https://github.com/cathyxl/MAgIC., Comment: work in progress

Recently, the development of large language models (LLMs) has been significantly enhanced the question answering and dialogue generation, and makes them become increasingly popular in current practical scenarios. While unlike the general dialogue system which emphasizes the semantic performance, the task-oriented dialogue (ToD) systems aim to achieve the dialogue goal efficiently and successfully in multiple turns. Unfortunately, existing LLM-induced ToD systems lack the direct reward toward the final goal and do not take account of the dialogue proactivity that can strengthen the dialogue efficiency. To fill these gaps, we introduce the ProToD (Proactively Goal-Driven LLM-Induced ToD) approach, which anticipates the future dialogue actions and incorporates the goal-oriented reward signal to enhance ToD systems. Additionally, we present a novel evaluation method that assesses ToD systems based on goal-driven dialogue simulations. This method allows us to gauge user satisfaction, system efficiency and successful rate while overcoming the limitations of current Information and Success metrics. Empirical experiments conducted on the MultiWoZ 2.1 dataset demonstrate that our model can achieve superior performance using only 10% of the data compared to previous end-to-end fully supervised models. This improvement is accompanied by enhanced user satisfaction and efficiency., Comment: 7 Pages

Empowering large language models to accurately express confidence in their answers is essential for trustworthy decision-making. Previous confidence elicitation methods, which primarily rely on white-box access to internal model information or model fine-tuning, have become less suitable for LLMs, especially closed-source commercial APIs. This leads to a growing need to explore the untapped area of black-box approaches for LLM uncertainty estimation. To better break down the problem, we define a systematic framework with three components: prompting strategies for eliciting verbalized confidence, sampling methods for generating multiple responses, and aggregation techniques for computing consistency. We then benchmark these methods on two key tasks-confidence calibration and failure prediction-across five types of datasets (e.g., commonsense and arithmetic reasoning) and five widely-used LLMs including GPT-4 and LLaMA 2 Chat. Our analysis uncovers several key insights: 1) LLMs, when verbalizing their confidence, tend to be overconfident, potentially imitating human patterns of expressing confidence. 2) As model capability scales up, both calibration and failure prediction performance improve. 3) Employing our proposed strategies, such as human-inspired prompts, consistency among multiple responses, and better aggregation strategies can help mitigate this overconfidence from various perspectives. 4) Comparisons with white-box methods indicate that while white-box methods perform better, the gap is narrow, e.g., 0.522 to 0.605 in AUROC. Despite these advancements, none of these techniques consistently outperform others, and all investigated methods struggle in challenging tasks, such as those requiring professional knowledge, indicating significant scope for improvement. We believe this study can serve as a strong baseline and provide insights for eliciting confidence in black-box LLMs., Comment: The paper is accepted by ICLR 2024. The code is publicly available at https://github.com/MiaoXiong2320/llm-uncertainty

Dialogue systems and large language models (LLMs) have gained considerable attention. However, the direct utilization of LLMs as task-oriented dialogue (TOD) models has been found to underperform compared to smaller task-specific models. Nonetheless, it is crucial to acknowledge the significant potential of LLMs and explore improved approaches for leveraging their impressive abilities. Motivated by the goal of leveraging LLMs, we propose an alternative approach called User-Guided Response Optimization (UGRO) to combine it with a smaller TOD model. This approach uses LLM as annotation-free user simulator to assess dialogue responses, combining them with smaller fine-tuned end-to-end TOD models. By utilizing the satisfaction feedback generated by LLMs, UGRO further optimizes the supervised fine-tuned TOD model. Specifically, the TOD model takes the dialogue history as input and, with the assistance of the user simulator's feedback, generates high-satisfaction responses that meet the user's requirements. Through empirical experiments on two TOD benchmarks, we validate the effectiveness of our method. The results demonstrate that our approach outperforms previous state-of-the-art (SOTA) results., Comment: Accepted by CIKM 2023

Controllable text generation is a challenging and meaningful field in natural language generation (NLG). Especially, poetry generation is a typical one with well-defined and strict conditions for text generation which is an ideal playground for the assessment of current methodologies. While prior works succeeded in controlling either semantic or metrical aspects of poetry generation, simultaneously addressing both remains a challenge. In this paper, we pioneer the use of the Diffusion model for generating sonnets and Chinese SongCi poetry to tackle such challenges. In terms of semantics, our PoetryDiffusion model, built upon the Diffusion model, generates entire sentences or poetry by comprehensively considering the entirety of sentence information. This approach enhances semantic expression, distinguishing it from autoregressive and large language models (LLMs). For metrical control, the separation feature of diffusion generation and its constraint control module enable us to flexibly incorporate a novel metrical controller to manipulate and evaluate metrics (format and rhythm). The denoising process in PoetryDiffusion allows for gradual enhancement of semantics and flexible integration of the metrical controller which can calculate and impose penalties on states that stray significantly from the target control distribution. Experimental results on two datasets demonstrate that our model outperforms existing models in automatic evaluation of semantic, metrical, and overall performance as well as human evaluation., Comment: Accepted by AAAI2024

Continual learning (CL) has attracted increasing attention in the recent past. It aims to mimic the human ability to learn new concepts without catastrophic forgetting. While existing CL methods accomplish this to some extent, they are still prone to semantic drift of the learned feature space. Foundation models, which are endowed with a robust feature representation, learned from very large datasets, provide an interesting substrate for the solution of the CL problem. Recent work has also shown that they can be adapted to specific tasks by prompt tuning techniques that leave the generality of the representation mostly unscathed. An open question is, however, how to learn both prompts that are task specific and prompts that are global, i.e. capture cross-task information. In this work, we propose the Prompt Of Prompts (POP) model, which addresses this goal by progressively learning a group of task-specified prompts and a group of global prompts, denoted as POP, to integrate information from the former. We show that a foundation model equipped with POP learning is able to outperform classic CL methods by a significant margin. Moreover, as prompt tuning only requires a small set of training samples, POP is able to perform CL in the few-shot setting, while still outperforming competing methods trained on the entire dataset.

Safety is a crucial property of every robotic platform: any control policy should always comply with actuator limits and avoid collisions with the environment and humans. In reinforcement learning, safety is even more fundamental for exploring an environment without causing any damage. While there are many proposed solutions to the safe exploration problem, only a few of them can deal with the complexity of the real world. This paper introduces a new formulation of safe exploration for reinforcement learning of various robotic tasks. Our approach applies to a wide class of robotic platforms and enforces safety even under complex collision constraints learned from data by exploring the tangent space of the constraint manifold. Our proposed approach achieves state-of-the-art performance in simulated high-dimensional and dynamic tasks while avoiding collisions with the environment. We show safe real-world deployment of our learned controller on a TIAGo++ robot, achieving remarkable performance in manipulation and human-robot interaction tasks., Comment: 6 pages

Qiongrong Huang,1,2 Xiuli Zhang,2 Zhiyuan Hu1– 4 1Fujian Provincial Key Laboratory of Brain Aging and Neurodegenerative Diseases, School of Basic Medical Sciences, Fujian Medical University, Fuzhou, 350108, Fujian Province, People’s Republic of China; 2CAS Key Laboratory of Standardization and Measurement for Nanotechnology, CAS Key Laboratory for Biomedical Effects of Nanomaterials and Nanosafety, CAS Center for Excellence in Nanoscience, National Center for Nanoscience and Technology of China, Beijing, 100190, People’s Republic of China; 3School of Nanoscience and Technology, Sino-Danish College, University of Chinese Academy of Sciences, Beijing, 100049, People’s Republic of China; 4School of Chemical Engineering and Pharmacy, Wuhan Institute of Technology, Wuhan, 430205, People’s Republic of ChinaCorrespondence: Zhiyuan HuFujian Provincial Key Laboratory of Brain Aging and Neurodegenerative Diseases, School of Basic Medical Sciences, Fujian Medical University, Fuzhou, 350108, Fujian Province, People’s Republic of ChinaTel/Fax +86 10 6265 2116Email huzy@nanoctr.cnXiuli ZhangNational Center for Nanoscience and Technology (NCNST),No.11 ZhongGuanCun BeiYiTiao, Beijing, 100190, People’s Republic of ChinaTel +010 8254 5752Email zhxiuli@gmail.comPurpose: Inflammatory bowel disease (IBD) is difficult to diagnose and classify. The purpose of this study is to establish an artificial intelligence model based on fecal multi-omics data for multi-classification diagnosis of IBD and its subtypes.Materials and Methods: A total of 299 clinical cohort studies were included in this study, including 86 healthy people, 140 CD patients and 73 UC patients. Based on the idea of hierarchical modeling for different groups, we model the total population and the groups with self-evaluation of “very well” and “slightly below par”, respectively. The original total features were fecal

Iman Rostami,1,2 Hamideh Rezvani Alanagh,2 Zhiyuan Hu,2,3 Sarah H Shahmoradian1 1Laboratory of Biomolecular Research, Department of Biology and Chemistry, Paul Scherrer Institute, Villigen, PSI 5232, Switzerland; 2CAS Key Laboratory of Standardization and Measurement for Nanotechnology, CAS Key Laboratory for Biomedical Effects of Nanomaterials and Nanosafety, CAS Center for Excellence in Nanoscience, National Center for Nanoscience and Technology of China, Beijing 100190, People’s Republic of China; 3Center for Neuroscience Research, School of Basic Medical Sciences, Fujian Medical University, Fuzhou, Fujian Province 350108, People’s Republic of ChinaCorrespondence: Iman RostamiPaul Scherrer Institute, OFLC/110, Forschungsstrasse 111, Villigen, PSI 5232, SwitzerlandTel +41 56 310 3309Email iman.rostami@psi.chAbstract: Nanomedicine is a medical application of biochemistry incorporated with materials chemistry at the scale of nanometer for the purpose of diagnosis, prevention, and treatment. New models and approaches are typically associated with nanomedicine for precise multifunctional diagnostic systems at molecular level. Hence, employing nanoparticles (NPs) has unveiled new opportunities for efficient therapies and remedy of difficult-to-cure diseases. Among all types of inorganic NPs, lanthanide-doped up-conversion nanoparticles (UCNPs) have shown excellent potential for biomedical applications, especially for multimodal bioimaging including fluorescence and electron microscopy. Association of these visualization techniques plus the capability for transporting biomaterials and drugs make them superior agents in the field of nanomedicine. Accordingly, in this review, we firstly presented a fundamental understanding of physical and optical properties of UCNPs and secondly, we illustrated some of the prominent associations with bioimaging, theranostics, cancer therapy, and optogenetics.Keywords: up-conversion nanoparticles, bioimaging, theranost

Virtually all of deep learning literature relies on the assumption of large amounts of available training data. Indeed, even the majority of few-shot learning methods rely on a large set of "base classes" for pretraining. This assumption, however, does not always hold. For some tasks, annotating a large number of classes can be infeasible, and even collecting the images themselves can be a challenge in some scenarios. In this paper, we study this problem and call it "Small Data" setting, in contrast to "Big Data". To unlock the full potential of small data, we propose to augment the models with annotations for other related tasks, thus increasing their generalization abilities. In particular, we use the richly annotated scene parsing dataset ADE20K to construct our realistic Long-tail Recognition with Diverse Supervision (LRDS) benchmark by splitting the object categories into head and tail based on their distribution. Following the standard few-shot learning protocol, we use the head classes for representation learning and the tail classes for evaluation. Moreover, we further subsample the head categories and images to generate two novel settings which we call "Scarce-Class" and "Scarce-Image", respectively corresponding to the shortage of samples for rare classes and training images. Finally, we analyze the effect of applying various additional supervision sources under the proposed settings. Our experiments demonstrate that densely labeling a small set of images can indeed largely remedy the small data constraints., Comment: Learning from Limited and Imperfect Data (L2ID) Workshop @ CVPR 2021

Online Social Networks (OSNs) evolve through two pervasive behaviors: follow and unfollow, which respectively signify relationship creation and relationship dissolution. Researches on social network evolution mainly focus on the follow behavior, while the unfollow behavior has largely been ignored. Mining unfollow behavior is challenging because user's decision on unfollow is not only affected by the simple combination of user's attributes like informativeness and reciprocity, but also affected by the complex interaction among them. Meanwhile, prior datasets seldom contain sufficient records for inferring such complex interaction. To address these issues, we first construct a large-scale real-world Weibo dataset, which records detailed post content and relationship dynamics of 1.8 million Chinese users. Next, we define user's attributes as two categories: spatial attributes (e.g., social role of user) and temporal attributes (e.g., post content of user). Leveraging the constructed dataset, we systematically study how the interaction effects between user's spatial and temporal attributes contribute to the unfollow behavior. Afterwards, we propose a novel unified model with heterogeneous information (UMHI) for unfollow prediction. Specifically, our UMHI model: 1) captures user's spatial attributes through social network structure; 2) infers user's temporal attributes through user-posted content and unfollow history; and 3) models the interaction between spatial and temporal attributes by the nonlinear MLP layers. Comprehensive evaluations on the constructed dataset demonstrate that the proposed UMHI model outperforms baseline methods by 16.44% on average in terms of precision. In addition, factor analyses verify that both spatial attributes and temporal attributes are essential for mining unfollow behavior., Comment: 8 pages, 7 figures, Accepted by AAAI 2020

The effectiveness of most cancer targeted therapies is short-lived. Tumors often develop resistance that might be overcome with drug combinations. However, the number of possible combinations is vast, necessitating data-driven approaches to find optimal patient-specific treatments. Here we report AstraZeneca's large drug combination dataset, consisting of 11,576 experiments from 910 combinations across 85 molecularly characterized cancer cell lines, and results of a DREAM Challenge to evaluate computational strategies for predicting synergistic drug pairs and biomarkers. 160 teams participated to provide a comprehensive methodological development and benchmarking. Winning methods incorporate prior knowledge of drug-target interactions. Synergy is predicted with an accuracy matching biological replicates for >60% of combinations. However, 20% of drug combinations are poorly predicted by all methods. Genomic rationale for synergy predictions are identified, including ADAM17 inhibitor antagonism when combined with PIK3CB/D inhibition contrasting to synergy when combined with other PI3K-pathway inhibitors in PIK3CA mutant cells.

ObjectiveTriple-negative breast cancer (TNBC) is a heterogeneous disease with poor prognosis. Circulating tumor cells (CTCs) are a promising predictor for breast cancer prognoses but their reliability regarding progression-free survival (PFS) is controversial. We aim to verify their predictive value in TNBC.MethodsIn present prospective cohort study, we used the Pep@MNPs method to enumerate CTCs in baseline blood samples from 75 patients with TNBC (taken at inclusion in this study) and analyzed correlations between CTC numbers and outcomes and other clinical parameters.ResultsMedian PFS was 6.0 (range: 1.0-25.0) months for the entire cohort, in whom we found no correlations between baseline CTC status and initial tumor stage (P=0.167), tumor grade (P=0.783) or histological type (P=0.084). However, among those getting first-line treatment, baseline CTC status was positively correlated with ratio of peripheral natural killer (NK) cells (P=0.032), presence of lung metastasis (P=0.034) and number of visceral metastatic site (P=0.037). Baseline CTC status was predictive for PFS in first-line TNBC (P=0.033), but not for the cohort as a whole (P=0.118). This prognostic limitation of CTC could be ameliorated by combining CTC and NK cell enumeration (P=0.049).ConclusionsBaseline CTC status was predictive of lung metastasis, peripheral NK cell ratio and PFS in TNBC patients undergoing first-line treatment. We have developed a combined CTC-NK enumeration strategy that allows us to predict PFS in TNBC without any preconditions.

Isoprene dominates global non-methane volatile organic compound emissions, and impacts tropospheric chemistry by influencing oxidants and aerosols. Isoprene emission rates vary over several orders of magnitude for different plants, and characterizing this immense biological chemodiversity is a challenge for estimating isoprene emission from tropical forests. Here we present the isoprene emission estimates from aircraft eddy covariance measurements over the Amazonian forest. We report isoprene emission rates that are three times higher than satellite top-down estimates and 35% higher than model predictions. The results reveal strong correlations between observed isoprene emission rates and terrain elevations, which are confirmed by similar correlations between satellite-derived isoprene emissions and terrain elevations. We propose that the elevational gradient in the Amazonian forest isoprene emission capacity is determined by plant species distributions and can substantially explain isoprene emission variability in tropical forests, and use a model to demonstrate the resulting impacts on regional air quality.

Minzhi Zhao,* Haiyun Li,* Yan Ma, He Gong, Shu Yang, Qiaojun Fang, Zhiyuan Hu Chinese Academy of Sciences Key Laboratory for Biomedical Effects of Nanomaterials and Nanosafety, CAS Center for Excellence in Nanoscience, National Center for Nanoscience and Technology, Beijing, People’s Republic of China *These authors contributed equally to this work Abstract: Abraxane (Abr), a US Food and Drug Administration-approved albumin-bound nanoparticle applied for the treatment of non-small-cell lung cancer, has been reported to be more effective than paclitaxel (PTX). To further understand the molecular mechanisms that produce this superior drug efficacy of Abr, a quantitative proteomic approach has been applied to investigate the global protein expression profiles of lung cancer cell A549 treated with Abr and PTX. Only one protein, namely, glucosamine 6-phosphate N-acetyltransferase 1 (GNA1), showed significant differential expression (P<0.05) in the cutoff of 2.0 fold, suggesting that Abr can be used safely as a substitute for PTX. GNA1 is a key enzyme in the biosynthesis of uridine diphosphate-N-acetylglucosamine, which is an important donor substrate for N-linked glycosylation and has several important functions such as embryonic development and growth. Albumin plays a major role in the regulation of this protein. In summary, this study first shows that the superior drug effect of Abr is mainly due to the downregulation of GNA1, which causes proliferative delay and cell adhesion defect. It is also noteworthy that the deficiency of GNA1 might reduce insulin secretion which correlates with type 2 diabetes. Keywords: quantitative proteomics, nano-drug, drug efficacy, lung cancer, molecular mechanisms, abraxane

Biogenic volatile organic compounds (BVOCs) emitted from terrestrial ecosystems play an important role in atmospheric chemistry and global climate feedbacks. The immense biological and chemical diversity of BVOC is a challenge for the numerical modeling of BVOC emissions, especially for tropical forests. The fast response airborne PTR-MS measurements of BVOC mixing ratios during the GoAmazon2014/5 campaign provide an opportunity to estimate BVOC emissions from tropical forest landscapes using eddy covariance and mixed layer variance techniques. The average isoprene emissions based on airborne measurements are 6.16±4.62 mg m-2 h-1 during wet season and 12.89±8.93 mg m-2 h-1 during dry season. These observations are compared with the spatial and temporal distributions of BVOC emissions estimated using the Model of Emissions of Gases and Aerosols from Nature (MEGAN) within the framework of the Community Land Model (CLM). By comparing the emissions estimated from airborne observations with that from MEGAN simulations, the predictive capability for diverse tropical land cover types are evaluated and an approach for improving model estimates is outlined. Satellite observations are also applied to investigate the environmental driving variables for BVOC emissions. OH concentrations are calculated from the observed BVOC mixing ratios by using the mixed layer mass balance technique and by the gradient in measured eddy covariance fluxes. The calculated values are in general agreement with OH CIMS observations at the T3 ground site near Manacaparu.

We aimed to develop strategies for performing quantitative analyses of organ-specific blood biomarkers by: 1) developing better software for analyzing dynamically changing transcriptomes; 2) analyzing blood and multiple organs/tissues of animals from three inbred mouse strains exposed to the toxins acetaminophen and carbon tetrachloride for transcriptomes, proteins and miRNA biomarkers.; 3) establishing MRM mass spectrometry assays for at least 25 liver-specific blood proteins based on the acetaminophen, CCL4, and other model systems of interest; 4) analyzing blood and tissue of animals from three inbred mouse strains exposed to the toxins acetaminophen and carbon tetrachloride for protein biomarkers using proteomics technologies, including MRM; 5) Analyzing time course experiments of rat tissues and blood exposed to VX; and 6) Developing new technologies for developing protein-capture agents and the analyses of single protein molecules. Although discovery and validation of organ-specific biomarkers are challenging from a technical perspective, we made considerable progress towards achieving our stated aims.

Weihua Tang1, Zhiyuan Hu2, Hind Muallem1, Margaret L Gulley1,21Department of Pathology and Laboratory Medicine, 2Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, North Carolina, NC, USAAbstract: RNA profiling is increasingly used to predict drug response, dose, or toxicity based on analysis of drug pharmacokinetic or pharmacodynamic pathways. Before implementing multiplexed RNA arrays in clinical practice, validation studies are carried out to demonstrate sufficient evidence of analytic and clinical performance, and to establish an assay protocol with quality assurance measures. Pathologists assure quality by selecting input tissue and by interpreting results in the context of the input tissue as well as the technologies that were used and the clinical setting in which the test was ordered. A strength of RNA profiling is the array-based measurement of tens to thousands of RNAs at once, including redundant tests for critical analytes or pathways to promote confidence in test results. Instrument and reagent manufacturers are crucial for supplying reliable components of the test system. Strategies for quality assurance include careful attention to RNA preservation and quality checks at pertinent steps in the assay protocol, beginning with specimen collection and proceeding through the various phases of transport, processing, storage, analysis, interpretation, and reporting. Specimen quality is checked by probing housekeeping transcripts, while spiked and exogenous controls serve as a check on analytic performance of the test system. Software is required to manipulate abundant array data and present it for interpretation by a laboratory physician who reports results in a manner facilitating therapeutic decision-making. Maintenance of the assay requires periodic documentation of personnel competency and laboratory proficiency. These strategies are shepherding genomic arrays into clinical settings to provide added value to patients and

BackgroundAlthough numerous mouse models of breast carcinomas have been developed, we do not know the extent to which any faithfully represent clinically significant human phenotypes. To address this need, we characterized mammary tumor gene expression profiles from 13 different murine models using DNA microarrays and compared the resulting data to those from human breast tumors.ResultsUnsupervised hierarchical clustering analysis showed that six models (TgWAP-Myc, TgMMTV-Neu, TgMMTV-PyMT, TgWAP-Int3, TgWAP-Tag, and TgC3(1)-Tag) yielded tumors with distinctive and homogeneous expression patterns within each strain. However, in each of four other models (TgWAP-T121, TgMMTV-Wnt1, Brca1Co/Co;TgMMTV-Cre;p53+/- and DMBA-induced), tumors with a variety of histologies and expression profiles developed. In many models, similarities to human breast tumors were recognized, including proliferation and human breast tumor subtype signatures. Significantly, tumors of several models displayed characteristics of human basal-like breast tumors, including two models with induced Brca1 deficiencies. Tumors of other murine models shared features and trended towards significance of gene enrichment with human luminal tumors; however, these murine tumors lacked expression of estrogen receptor (ER) and ER-regulated genes. TgMMTV-Neu tumors did not have a significant gene overlap with the human HER2+/ER- subtype and were more similar to human luminal tumors.ConclusionMany of the defining characteristics of human subtypes were conserved among the mouse models. Although no single mouse model recapitulated all the expression features of a given human subtype, these shared expression features provide a common framework for an improved integration of murine mammary tumor models with human breast tumors.

BACKGROUND: Although numerous mouse models of breast carcinomas have been developed, we do not know the extent to which any faithfully represent clinically significant human phenotypes. To address this need, we characterized mammary tumor gene expression profiles from 13 different murine models using DNA microarrays and compared the resulting data to those from human breast tumors. RESULTS: Unsupervised hierarchical clustering analysis showed that six models (TgWAP-Myc, TgMMTV-Neu, TgMMTV-PyMT, TgWAP-Int3, TgWAP-Tag, and TgC3(1)-Tag) yielded tumors with distinctive and homogeneous expression patterns within each strain. However, in each of four other models (TgWAP-T121, TgMMTV-Wnt1, Brca1Co/Co;TgMMTV-Cre;p53+/- and DMBA-induced), tumors with a variety of histologies and expression profiles developed. In many models, similarities to human breast tumors were recognized, including proliferation and human breast tumor subtype signatures. Significantly, tumors of several models displayed characteristics of human basal-like breast tumors, including two models with induced Brca1 deficiencies. Tumors of other murine models shared features and trended towards significance of gene enrichment with human luminal tumors; however, these murine tumors lacked expression of estrogen receptor (ER) and ER-regulated genes. TgMMTV-Neu tumors did not have a significant gene overlap with the human HER2+/ER- subtype and were more similar to human luminal tumors. CONCLUSION: Many of the defining characteristics of human subtypes were conserved among the mouse models. Although no single mouse model recapitulated all the expression features of a given human subtype, these shared expression features provide a common framework for an improved integration of murine mammary tumor models with human breast tumors.

INTRODUCTION: Predicting the clinical course of breast cancer is often difficult because it is a diverse disease comprised of many biological subtypes. Gene expression profiling by microarray analysis has identified breast cancer signatures that are important for prognosis and treatment. In the current article, we use microarray analysis and a real-time quantitative reverse-transcription (qRT)-PCR assay to risk-stratify breast cancers based on biological 'intrinsic' subtypes and proliferation. METHODS: Gene sets were selected from microarray data to assess proliferation and to classify breast cancers into four different molecular subtypes, designated Luminal, Normal-like, HER2+/ER-, and Basal-like. One-hundred and twenty-three breast samples (117 invasive carcinomas, one fibroadenoma and five normal tissues) and three breast cancer cell lines were prospectively analyzed using a microarray (Agilent) and a qRT-PCR assay comprised of 53 genes. Biological subtypes were assigned from the microarray and qRT-PCR data by hierarchical clustering. A proliferation signature was used as a single meta-gene (log2 average of 14 genes) to predict outcome within the context of estrogen receptor status and biological 'intrinsic' subtype. RESULTS: We found that the qRT-PCR assay could determine the intrinsic subtype (93% concordance with microarray-based assignments) and that the intrinsic subtypes were predictive of outcome. The proliferation meta-gene provided additional prognostic information for patients with the Luminal subtype (P = 0.0012), and for patients with estrogen receptor-positive tumors (P = 3.4 x 10-6). High proliferation in the Luminal subtype conferred a 19-fold relative risk of relapse (confidence interval = 95%) compared with Luminal tumors with low proliferation. CONCLUSION: A real-time qRT-PCR assay can recapitulate microarray classifications of breast cancer and can risk-stratify patients using the intrinsic subtype and proliferation. The proliferation meta-gene

Background Validation of a novel gene expression signature in independent data sets is a critical step in the development of a clinically useful test for cancer patient risk-stratification. However, validation is often unconvincing because the size of the test set is typically small. To overcome this problem we used publicly available breast cancer gene expression data sets and a novel approach to data fusion, in order to validate a new breast tumor intrinsic list. Results A 105-tumor training set containing 26 sample pairs was used to derive a new breast tumor intrinsic gene list. This intrinsic list contained 1300 genes and a proliferation signature that was not present in previous breast intrinsic gene sets. We tested this list as a survival predictor on a data set of 311 tumors compiled from three independent microarray studies that were fused into a single data set using Distance Weighted Discrimination. When the new intrinsic gene set was used to hierarchically cluster this combined test set, tumors were grouped into LumA, LumB, Basal-like, HER2+/ER-, and Normal Breast-like tumor subtypes that we demonstrated in previous datasets. These subtypes were associated with significant differences in Relapse-Free and Overall Survival. Multivariate Cox analysis of the combined test set showed that the intrinsic subtype classifications added significant prognostic information that was independent of standard clinical predictors. From the combined test set, we developed an objective and unchanging classifier based upon five intrinsic subtype mean expression profiles (i.e. centroids), which is designed for single sample predictions (SSP). The SSP approach was applied to two additional independent data sets and consistently predicted survival in both systemically treated and untreated patient groups. Conclusion This study validates the "breast tumor intrinsic" subtype classification as an objective means of tumor classification that should be translated into a clinical ass