All models can be trained entirely on CPUs on consumer grade Laptop machines within minutes orhours. Execution times per epoch for the single-cell data with 529 features are as follows: Base=0.9, Centering the first frame: For golfing and waving, the root point of the first frame is movedtotheorigin(0,0,0). To map putative transcription factor (TF) and target gene relationships, we use as a reference a regulatory network generated using the gene expression and chromatin accessibility features 15 available inthehuman immune cells dataset. Ourruleforsuccessfully mapping aTFtoatargetgene through achromatin peak isthatall TF, chromatin peak, and target gene, have to be simultaneously in the list of features selected in therank_genes_groupsfunction for cell type of interest, and there haveto be TF motifs linked to that transcription factor in the chromatin peak.

While different types ofsparsity havebeen proposed toimproverobustness, the generalization properties ofNODEsfordynamical systemsbeyondtheobserved dataareunderexplored. Wesystematically studytheinfluenceofweight andfeature sparsity on forecasting as well as on identifying the underlying dynamical laws.

DNA, present in every cell, stores the complete set of instructions essential for life.

Effective DNALMs must, at minimum, achieve three key objectives: 1. Learn representations that can accurately distinguish different types of functional elements.

Whole-genome sequencing (WGS) has revealed numerous non-coding short variants whose functional impacts remain poorly understood. Despite recent advances in deep-learning genomic approaches, accurately predicting and prioritizing clinically relevant mutations in gene regulatory regions remains a major challenge. Here we introduce Deep VRegulome, a deep-learning method for prediction and interpretation of functionally disruptive variants in the human regulome, which combines 700 DNABERT fine-tuned models, trained on vast amounts of ENCODE gene regulatory regions, with variant scoring, motif analysis, attention-based visualization, and survival analysis. We showcase its application on TCGA glioblastoma WGS dataset in prioritizing survival-associated mutations and regulatory regions. The analysis identified 572 splice-disrupting and 9,837 transcription-factor binding site altering mutations occurring in greater than 10% of glioblastoma samples. Survival analysis linked 1352 mutations and 563 disrupted regulatory regions to patient outcomes, enabling stratification via non-coding mutation signatures. All the code, fine-tuned models, and an interactive data portal are publicly available.

Recent reasoning-focused language models achieve high accuracy by generating lengthy intermediate reasoning paths before producing final answers. While this approach is effective in solving problems that require logical thinking, long reasoning paths significantly increase memory usage and reduce throughput of token generation, limiting the practical deployment of such models. We propose Reasoning Path Compression (RPC), a training-free method that accelerates inference by leveraging the semantic sparsity of reasoning paths. RPC periodically compresses the KV cache by retaining cache entries that receive high importance score, which are computed using a selector window composed of recently generated queries. Experiments show that RPC improves generation throughput of QwQ-32B by up to 1.60$\times$ compared to the inference with full KV cache, with an accuracy drop of 1.2\% on the AIME 2024 benchmark. Our findings demonstrate that semantic sparsity in reasoning traces can be effectively exploited for compression, offering a practical path toward efficient deployment of reasoning LLMs. Our code is available at https://github.com/jiwonsong-dev/ReasoningPathCompression.

Effective DNALMs must, at minimum, achieve three key objectives: 1. Learn representations that can accurately distinguish different types of functional elements.

We provide a detailed proof for Theorem 1. We provide an alternative proof for identifiability of Poisson BN. I (G D), where the last equality holds because the integrand is the kernel of a beta distribution. The scRNA-seq experiments were performed on five mice with AhR knockout targeted to intestinal stem cells. On average each mouse contributed 6,000 cells.