01 / RESEARCH APPROACH

研究方向与代表工作Research directions & selected studies

My work begins with ecological variation in microbial communities and asks how robust patterns can be related to human phenotypes and brain function—without assuming causality from association.从微生物群落的生态差异出发,研究可复现的模式如何与临床表型及脑功能关联,同时明确统计关联与因果机制之间的距离。

这里介绍研究问题、分析框架与三项代表工作。每项研究都保留论文原文、图件出处及相关代码入口。Explore the research questions, analytical framework and three selected studies, with original articles, figure sources and relevant code.

RESEARCH FRAMEWORK

Questions before methods.先有科学问题,再选择方法。

01 / MAP

Ecological structure微生物生态结构

Community composition, variability and ecological factors across cohorts and environments.比较不同队列和环境中的群落组成、变异性与连续生态因子。

02 / LINK

Clinical heterogeneity临床异质性

Evaluate associations with symptoms, cognition and phenotypic diversity.探索微生物生态特征与症状、认知和临床表型差异的统计关联。

03 / CONTEXT

Brain signatures脑功能与空间特征

Integrate neuroimaging and spatial reference maps with transparent evidence boundaries.整合脑影像与空间参考图谱,明确跨数据集解释所受的证据限制。

04 / TEST

Robustness & validation稳健性与独立验证

External data, sensitivity analyses and limitations as part of the result.将外部数据验证、敏感性分析及局限性纳入研究结果本身。

SELECTED STUDIES

Evidence in context.让证据回到科学问题中。

012026 / ADVANCED SCIENCE

A continuous picture of global gut ecology.从全球宏基因组识别连续生态结构。

Wiredancer applied a similarity-constrained factor model to 20,178 metagenomes from 36 countries and over 50 disease states. The study reported three overlapping ecological factors and tested their reproducibility across cohorts.Wiredancer通过相似性约束的生态因子建模分析20,178份来自36国、50余疾病状态的宏基因组数据,识别重叠的连续微生物生态因子,并检验其跨队列复现性。

Zhu et al. · Advanced Science · 2026 · Fig. 1
CC BY 4.0 · 原图,未作修改Original figure, unmodified
20,178 metagenomes36 countries
022026 / RESEARCH

A cohort designed for the brain–gut question.面向脑肠关联的前瞻性队列。

The Brain–Gut Health Initiative describes a platform for jointly studying microbiome characteristics, psychiatric symptoms, cognitive assessments and neuroimaging, with emphasis on participant-level data alignment and clinically meaningful heterogeneity.脑肠健康计划结合微生物组、精神疾病症状、认知评估与脑影像,为探索个体层面的跨系统关联及临床异质性提供队列基础。

Wu et al. · Research · 2026 · Fig. 1
CC BY 4.0 · 原图,未作修改Original figure, unmodified
Clinical phenotypesMultimodal framework
032026 / JOURNAL OF AFFECTIVE DISORDERS

Beyond bacteria in depression.抑郁症微生物组研究不止于细菌。

A metagenomic study of first-episode drug-naïve depression profiled bacteria, archaea, viruses and microbial eukaryotes, and assessed clinical associations plus the exploratory discriminatory potential of combined features.在首发未用药抑郁症队列中联合分析细菌、古菌、病毒和微型真核生物,评估其临床相关性及联合特征的探索性区分潜力。

Four microbial kingdomsClinical associations
RESEARCH PRINCIPLE

A good model should travel beyond its training set.好的模型,应当走出训练数据。

External validation, confounding, missingness and measurement differences are not an appendix to biological discovery. They determine what the evidence can support.独立验证、混杂因素、缺失值与测量差异并不是生物学发现之后才考虑的附加问题,而是决定研究证据能够支持何种结论的关键。