Search for: All records

Large Language Models (LLMs) have catalyzed vibe coding, where users leverage LLMs to generate and iteratively refine code through natural language interactions until it passes their vibe check. Vibe check reflects human preference and goes beyond functionality: the solution should feel right, read cleanly, preserve intent, and remain correct. However, current code evaluation remains anchored to pass@k and captures only functional correctness, overlooking non-functional instructions that users routinely apply. In this paper, we hypothesize that instruction following is the missing piece underlying vibe check besides functional correctness. To quantify models' code instruction-following capabilities with measurable signals, we present VeriCode, a taxonomy of 30 verifiable code instructions together with deterministic verifiers. We use the taxonomy to augment established evaluation suites, resulting in SWE-IF, a testbed to assess both instruction following and functional correctness. Evaluating 31 LLMs, we show that even the strongest models struggle to comply with multiple instructions and exhibit functional regression. Most importantly, a composite score of functional correctness and instruction following correlates best with human preference, with instruction following emerging as the primary differentiator among LLMs. Our code, data, and taxonomy are available at https://github.com/maszhongming/SWE-IF. 

Abstract In single cell biology, the complexity of tissues may hinder lineage cell mapping or tumor microenvironment decomposition, requiring digital dissociation of bulk tissues. Many deconvolution methods focus on transcriptomic assay, not easily applicable to other omics due to ambiguous cell markers and reference-to-target difference. Here, we present MOADE, a multimodal autoencoder pipeline linking multi-dimensional features to jointly predict personalized multi-omic profiles and cellular compositions, using pseudo-bulk data constructed by internal non-transcriptomic reference and external scRNA-seq data. MOADE is evaluated through rigorous simulation experiments and real multi-omic data from multiple tissue types, outperforming nine deconvolution pipelines with superior generalizability and fidelity. 

Alternative cleavage and polyadenylation within introns (intronic APA) generate shorter mRNA isoforms; however, their physiological significance remains elusive. In this study, we developed a comprehensive workflow to analyze intronic APA profiles using the mammalian target of rapamycin (mTOR)-regulated transcriptome as a model system. Our investigation revealed two contrasting effects within the transcriptome in response to fluctuations in cellular mTOR activity: an increase in intronic APA for a subset of genes and a decrease for another subset of genes. The application of this workflow to RNA-seq data from The Cancer Genome Atlas demonstrated that this dichotomous intronic APA pattern is a consistent feature in transcriptomes across both normal tissues and various cancer types. Notably, our analyses of protein length changes resulting from intronic APA events revealed two distinct phenomena in proteome programming: a loss of functional domains due to significant changes in protein length or minimal alterations in C-terminal protein sequences within unstructured regions. Focusing on conserved intronic APA events across 10 different cancer types highlighted the prevalence of the latter cases in cancer transcriptomes, whereas the former cases were relatively enriched in normal tissue transcriptomes. These observations suggest potential, yet distinct, roles for intronic APA events during pathogenic processes and emphasize the abundance of protein isoforms with similar lengths in the cancer proteome. Furthermore, our investigation into the isoform-specific functions of JMJD6 intronic APA events supported the hypothesis that alterations in unstructured C-terminal protein regions lead to functional differences. Collectively, our findings underscore intronic APA events as a discrete molecular signature present in both normal tissues and cancer transcriptomes, highlighting the contribution of APA to the multifaceted functionality of the cancer proteome. 

Alternative cleavage and polyadenylation within introns (intronic APA) generate shorter mRNA isoforms; however, their physiological significance remains elusive. In this study, we developed a comprehensive workflow to analyze intronic APA profiles using the mammalian target of rapamycin (mTOR)-regulated transcriptome as a model system. Our investigation revealed two contrasting effects within the transcriptome in response to fluctuations in cellular mTOR activity: an increase in intronic APA for a subset of genes and a decrease for another subset of genes. The application of this workflow to RNA-seq data from The Cancer Genome Atlas demonstrated that this dichotomous intronic APA pattern is a consistent feature in transcriptomes across both normal tissues and various cancer types. Notably, our analyses of protein length changes resulting from intronic APA events revealed two distinct phenomena in proteome programming: a loss of functional domains due to significant changes in protein length or minimal alterations in C- terminal protein sequences within unstructured regions. Focusing on conserved intronic APA events across 10 different cancer types highlighted the prevalence of the latter cases in cancer transcriptomes, whereas the former cases were relatively enriched in normal tissue transcriptomes. These observations suggest potential, yet distinct, roles for intronic APA events during pathogenic processes and emphasize the abundance of protein isoforms with similar lengths in the cancer proteome. Furthermore, our investigation into the isoform-specific functions of JMJD6 intronic APA events supported the hypothesis that alterations in unstructured C-terminal protein regions lead to functional differences. Collectively, our findings underscore intronic APA events as a discrete molecular signature present in both normal tissues and cancer transcriptomes, highlighting the contribution of APA to the multifaceted functionality of the cancer proteome. 

The goal of the crime forecasting problem is to predict different types of crimes for each geographical region (like a neighborhood or censor tract) in the near future. Since nearby regions usually have similar socioeconomic characteristics which indicate similar crime patterns, recent state-of-the-art solutions constructed a distance-based region graph and utilized Graph Neural Network (GNN) techniques for crime forecasting, because the GNN techniques could effectively exploit the latent relationships between neighboring region nodes in the graph if the edges reveal high dependency or correlation. However, this distance-based pre-defined graph can not fully capture crime correlation between regions that are far from each other but share similar crime patterns. Hence, to make a more accurate crime prediction, the main challenge is to learn a better graph that reveals the dependencies between regions in crime occurrences and meanwhile captures the temporal patterns from historical crime records. To address these challenges, we propose an end-to-end graph convolutional recurrent network called HAGEN with several novel designs for crime prediction. Specifically, our framework could jointly capture the crime correlation between regions and the temporal crime dynamics by combining an adaptive region graph learning module with the Diffusion Convolution Gated Recurrent Unit (DCGRU). Based on the homophily assumption of GNN (i.e., graph convolution works better where neighboring nodes share the same label), we propose a homophily-aware constraint to regularize the optimization of the region graph so that neighboring region nodes on the learned graph share similar crime patterns, thus fitting the mechanism of diffusion convolution. Empirical experiments and comprehensive analysis on two real-world datasets showcase the effectiveness of HAGEN. 

Abstract Motivation Accurate disease phenotype prediction plays an important role in the treatment of heterogeneous diseases like cancer in the era of precision medicine. With the advent of high throughput technologies, more comprehensive multi-omics data is now available that can effectively link the genotype to phenotype. However, the interactive relation of multi-omics datasets makes it particularly challenging to incorporate different biological layers to discover the coherent biological signatures and predict phenotypic outcomes. In this study, we introduce omicsGAN, a generative adversarial network model to integrate two omics data and their interaction network. The model captures information from the interaction network as well as the two omics datasets and fuse them to generate synthetic data with better predictive signals. Results Large-scale experiments on The Cancer Genome Atlas breast cancer, lung cancer and ovarian cancer datasets validate that (i) the model can effectively integrate two omics data (e.g. mRNA and microRNA expression data) and their interaction network (e.g. microRNA-mRNA interaction network). The synthetic omics data generated by the proposed model has a better performance on cancer outcome classification and patients survival prediction compared to original omics datasets. (ii) The integrity of the interaction network plays a vital role in the generation of synthetic data with higher predictive quality. Using a random interaction network does not allow the framework to learn meaningful information from the omics datasets; therefore, results in synthetic data with weaker predictive signals. Availability and implementation Source code is available at: https://github.com/CompbioLabUCF/omicsGAN. Supplementary information Supplementary data are available at Bioinformatics online. 

Microbes and viruses are known to alter host transcriptomes by means of infection. In light of recent challenges posed by the COVID-19 pandemic, a deeper understanding of the disease at the transcriptome level is needed. However, research about transcriptome reprogramming by post-transcriptional regulation is very limited. In this study, computational methods developed by our lab were applied to RNA-seq data to detect transcript variants (i.e., alternative splicing (AS) and alternative polyadenylation (APA) events). The RNA-seq data were obtained from a publicly available source, and they consist of mock-treated and SARS-CoV-2 infected (COVID-19) lung alveolar (A549) cells. Data analysis results show that more AS events are found in SARS-CoV-2 infected cells than in mock-treated cells, whereas fewer APA events are detected in SARS-CoV-2 infected cells. A combination of conventional differential gene expression analysis and transcript variants analysis revealed that most of the genes with transcript variants are not differentially expressed. This indicates that no strong correlation exists between differential gene expression and the AS/APA events in the mock-treated or SARS-CoV-2 infected samples. These genes with transcript variants can be applied as another layer of molecular signatures for COVID-19 studies. In addition, the transcript variants are enriched in important biological pathways that were not detected in the studies that only focused on differential gene expression analysis. Therefore, the pathways may lead to new molecular mechanisms of SARS-CoV-2 pathogenesis.