Yangyang Wu

Open access Sep 2026

iPscDB: a comprehensive platform for plant single-cell transcriptomic data integration and analysis.

The rapid advancement of single-cell technologies has significantly enhanced our ability to investigate cellular heterogeneity within plant tissues. However, deciphering these intricate cellular landscapes requires processing high-dimensional gene expression matrices and integrating diverse datasets to enable accurate marker selection, cell identification, and other complex computational operations. These processes typically require broad programming expertise, posing a challenge for researchers with a limited computational background. To address this, we present integrated Plant single-cell Database (iPscDB), an integrated and multifunctional platform that facilitates the integration and analysis of plant single-cell data. iPscDB combines 4 688 428 cells and 288 139 curated cell markers derived from 946 experiments across 38 plant species. Wherever raw data were available, datasets were reprocessed through a single uniform pipeline, and both integration quality and automated cell-type annotation were benchmarked quantitatively. The platform introduces a Marker Confidence Level scheme that grades each cell-type marker by the strength and independence of its supporting evidence (from manually curated classic markers to database-derived associations), allowing users to judge marker reliability directly. The platform also provides a user-friendly online analysis pipeline and modules capable of processing raw FASTQ files or Cell Ranger-processed files. Users can configure parameters via an intuitive interface and utilize an integrated image editor to customize visualization outputs. Additionally, iPscDB supports various analyses, including cross-species gene expression, electronic Single-Cell Pictograph, and developmental trajectory. By streamlining the complex workflows of single-cell transcriptomics, iPscDB offers a practical and accessible resource for researchers with diverse technical backgrounds. iPscDB is accessible at https://www.tobaccodb.org/ipscdb/homePage.

Peng Lu, Jingjing Jin, Jie-Meng Tao et al. · Nucleic Acids Research · 0 citations

Preprint Aug 2026

Taming the Implicit: Dual-Channel Risk-Aware Reinforcement Fine-Tuning for Continual Multimodal Post-Training

Reinforcement fine-tuning (RFT) is widely believed to inherently resist catastrophic forgetting in continual post-training of multimodal large language models. Under pronounced task distributional shifts, however, forgetting across representative RFT algorithms escalates sharply. This stems from the implicit reward-variance regularization inherent to RFT, which proves incapable of suppressing uncontrolled optimization risk. We propose Risk-Aware Policy Optimization (RAPO), the first dual-channel framework for explicit risk governance in continual RFT. On the policy channel, Risk-Aware Policy Scaling adaptively calibrates per-sample update magnitude via rollout reliability and Fisher-inspired local predictive sensitivity; on the data channel, Risk-Aware Dynamic Bucket Sampling reorganizes training batches through dynamic risk stratification, steering optimization toward informative yet stable samples. As a plug-and-play strategy requiring no cross-task memory, RAPO generalizes to any RFT algorithm without modification. On the public MLLM-CL benchmark, RAPO reduces final forgetting by 79.8% relative to its RLOO backbone while retaining new-task competitiveness.

Yibei Liu, Jiajun Chen, Qianle Zhang et al. · 0 citations

Preprint Aug 2026

Contrastive Mixed Prompt Learning for Incomplete Multimodal Sentiment Analysis with Unseen Modality Combination

This paper introduces the problem of Incomplete Multimodal Sentiment Analysis with Unseen Modality Combinations (IMSAUMC), aiming to enhance model generalization for unseen modality combinations, and introduces a label-guided contrastive feature learning mechanism to learn robust and discriminative cross-modal representations.

Kaixin Xu, Nai-Jing Liu, Yu-Ri Kang et al. · 0 citations