Author

Yang Liu

View on Semantic Scholar

2 papers indexed here

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Open access Jul 2026

A chromosome-level, gap-free genome assembly of the fish parasite Trypanosoma larimichthysi from large yellow croaker.

Marine fish trypanosomes are widespread hemoparasites that pose significant threats to wild and farmed teleosts, yet they remain genomically underrepresented compared to their mammalian-infecting counterparts. Here, we present the first chromosome-level, gap-free genome assembly of Trypanosoma larimichthysi, a recently described species causing severe trypanosomiasis outbreaks in the economically important large yellow croaker (Larimichthys crocea) along the Chinese coast. The assembly, generated using PacBio HiFi long-read sequencing combined with Hi-C chromatin conformation capture, spans 51.04 Mb across exactly 35 pseudochromosomes, with a contig N50 of 1.43 Mb and 97.96% of sequences anchored. Exceptional completeness is evidenced by telomere-to-telomere resolution for 25 chromosomes (58 telomeric loci captured in total), a 99.99% HiFi read mapping rate, and > 99% BUSCO completeness. The genome encodes 10,172 protein-coding genes, with repetitive sequences comprising 50.33%, dominated by retrotransposons including LINE and LTR elements. Comparative genomic analyses confirm the phylogenetic placement of T. larimichthysi within Trypanosoma and reveal lineage-specific gene family expansions potentially linked to host adaptation and pathogenicity. This reference-grade genome fills a critical gap in aquatic trypanosomatid genomics and provides a valuable resource for investigating parasite evolution, host-parasite interactions, antigenic variation mechanisms, and disease management strategies in mariculture.

Xin Lv, Lanying Liu, Zhideng Zhang et al. · Scientific Data · 0 citations

Preprint Jul 2026

Understanding and Mitigating the Video-Action Generalization Gap via Temporal Ratio

Generative video foundation models exhibit strong compositional priors, yet world-action models (WAMs) and video-action models (VAMs) often lose these priors after finetuning on robotic action data. We refer to this discrepancy as the video-action generalization gap. In this paper, we systematically investigate this gap by evaluating a comprehensive design space of VAMs, demonstrating that standard design choices yield no emergent explanation pattern. To explain this behavior, we introduce the Temporal Ratio (TR), an attention-based measure of how strongly the action head relies on future latent rollouts relative to the anchored current frame. TR has two key properties: first, a model's structural reliance on future-predictive latents, measured via TR, acts as a predictor of its compositional generalization capacity; second, it natively fluctuates based on task phase, shifting attention to future frames during planning and reverting to the present frame for precise manipulation. Finally, based on these findings, we propose an inference-time adaptive guidance method, which exploits this intrinsic feature attention pattern to dynamically amplify compositional video conditioning signals precisely when the policy relies on future rollouts. Evaluated on the LIBERO benchmark and real-world tasks, our approach mitigates the OOD-ID compositional generalization gap. More details: https://umishra.me/temporal-ratio/

U. Mishra, Yongxin Chen, Danfei Xu et al. · 1 citation