Biology subjects

Tackmann, J.

Publications and source records attributed to Tackmann, J..

8 recordsLinked to original sources

bioRxiv2026

Susagi: A Microbiome World Model

MotivationAccurately modelling how microbial communities assemble and change across hosts and environments is essential for analysis and intervention. Typical pipelines capture limited generalisable structure and often depend on fixed ecological unit definitions. ResultsWe present Susagi (Set Unsupervised Assessment of Genetic Imposters), a permutation-invariant denoising transformer that operates directly on sets of bacterial SSU rRNA gene embeddings to learn a member-level stability function. The model was trained on [~] 2 x 106 bacterial community samples. We show that it reliably predicts community composition dynamics in a zero-shot (no training) setting, demonstrated here across three challenging microbiomes for which traditional ML methods do not exceed random expectation. The models stability scores capture biological structure: across datasets, higher scores are enriched for agricultural, cropland, and soil-associated habitats, consistent with microbial communities in these environments supporting positive diversity-stability relationships. The highest stability scores are only attained by communities with high Pielou evenness and large size, despite the fact that the model has never seen abundances suggesting it can recognise community dysbiosis from presence absence alone. Furthermore, they also track biological gradients such as subject age. Susagi is competitive with another Large Microbiome Model (Microbial General Model, MGM) on diverse classification tasks, without task-specific fine-tuning and with an increased parameter efficiency. Ultimately, our model will facilitate hypothesis generation for complex microbial processes, including deterministic assembly and microbial interactions, crucial for instance in the design of communities in silico. Availability and implementationEvaluation code and model weights can be found from https://github.com/the-puzzler/Microbiome-Modelling. Model weights can also be downloaded directly from https://huggingface.co/basilboy/microbiome-model. Interactive demo can be found here: https://huggingface.co/spaces/basilboy/microbiome-space.

systems biology↗

bioRxiv2025

The MicrobeAtlas database: Global trends and insights into Earth's microbial ecosystems

Environmental DNA sequencing has revolutionized our understanding of microbial diversity and ecology. Microbiomes have now been sequenced across the entire planet--from the deep subsurface to the mountain tops--covering a myriad of hosts, biomes, and conditions. Yet, the diversity of sequencing and processing strategies hampers universal insights. MicrobeAtlas unifies more than two million microbiome samples in a single resource, harmonized to facilitate discoveries across technologies. Communities are hierarchically quantified at adjustable SSU rRNA marker gene resolution and feature detailed metadata, including rich geographic information. Connections to genome, phenotype, and ecological resources enable multimodal insights. Microbial lineages can be reliably tracked across environments, including a long tail of rare, uncharacterized species. Recurring community structures and geographic preferences become apparent, and global, taxonomy-specific generalism trends emerge. With MicrobeAtlas (www.microbeatlas.org), both known and newly described species and communities can readily be placed into ecological context, taking full advantage of earlier work.

bioinformatics↗

bioRxiv2025

Enhanced semantic classification of microbiome sample origins using Large Language Models (LLMs)

Over the past decade, central sequence repositories have expanded significantly in size. This vast accumulation of data holds value and enables further studies, provided that the data entries are well annotated. However, the submitter-provided metadata of sequencing records can be of heterogeneous quality, presenting significant challenges for re-use. Here, we test to what extent large language models (LLMs) can be used to cost-effectively automate the re-annotation of sequencing records against a simplified classification scheme of broad ecological environments with relevance to microbiome studies, without retraining. We focused on sequencing samples taken from the environment, for which metadata is important. We employed OpenAI Generative Pretrained Transformer (GPT) models, and assessed scalability, time and cost-effectiveness, as well as performance against a diverse, hand-curated ground-truth benchmark with 1000 examples, that span a wide range of complexity in metadata interpretation. We observed that annotation performance markedly outperforms that of a baseline, manually curated, non-machine-learning keyword-based approach. Changing models (or model parameters) has only minor effects on performance, but prompts need to be carefully designed to match the task. We applied the optimized pipeline to more than 3.8 million sequencing records from the environment, providing coarse-grained yet standardized sampling site annotations covering the globe. Our work demonstrates the effective use of LLMs to simplify and standardize annotation from complex biological metadata.

bioinformatics↗

bioRxiv2024

Habitat filtering but not microbiota origin controls microbiome transplant outcomes in soil

Human activities cause global losses of soil microbiome diversity and functionality. Microbiota transplants offer a potential solution, but the factors influencing transplant success remain unclear. We investigated how microbiota origin affects microbiome mergers, hypothesizing that native strains through niche preference are better adapted to their habitat and will outcompete non-native ones. To test this, we contrasted transplants between soil microcosm-cultured topsoil or lakewater communities with a community of 21 soil bacteria (SynCom). In both cases, SynCom transplant increased resident productivity but permanently shifted compositions, although its abundance dropped from an initial 50-80% to <1% within two months. Both merged and non-merged communities resembled natural soil microbiota in comparisons with over 81,000 soil, sediment and lake compositional data. Our results show that habitat filtering and niche competition, more than microbiota origin, determine transplant outcomes. Despite the limited proliferation of SynCom transplants, their capacity to instill lasting community trajectory changes opens new paths for microbiome engineering. TEASEREven transiently present microbiota transplants can alter resident microbiome composition through processes governed by habitat filtering.

microbiology↗

bioRxiv2024

Community conservatism is widespread throughout microbial phyla and environments

Evolution gives rise to various long-term phenomena, including Phylogenetic signal, which describes the tendency of related biological taxa to resemble each other in morphology and function. Related taxa tend to also live in similar ecological niches - this trend is termed Niche conservatism. Both concepts are widely used to understand crucial aspects of evolu1on and specia1on and are well-established in animals and plants. The extension of these concepts to microorganisms is however challenging and thus far only assumed. Here, we hypothesize that two closely related microbial species, if indeed similar in both their morphology and their ecological niche, should be found in samples with similar community compositions. We propose "community conservatism" to refer to this, and leverage a database with millions of samples and hundreds of thousands of pairs of microbial taxa to assess their relatedness and the similarity of the communities they occupy. Our findings reveal that community conservatism can be observed globally in all environments and phyla tested, to varying extents. Ecologically specialized taxa show a stronger community conservatism signal than generalists, and the signal can be discerned over nearly all taxonomic ranks. Analyzing community conservatism shows promise to advance our understanding of evolution, speciation, and the mechanisms governing community assembly in microorganisms. Furthermore, we propose it can be used to reintegrate ecological parameters into Operational Taxonomic Unit delimitation.

ecology↗

bioRxiv2021

Reproducible Propagation of Species-Rich Soil Microbiomes Suggests Robust Underlying Deterministic Principles of Community Formation

Microbiomes are typically characterised by high species diversity but it is poorly understood how such system-level complexity can be generated and propagated. Here, we used soils as a relevant model to study microbiome development. Despite inherent stochastic variation in manipulating species-rich communities, both laboratory-mixed medium complexity (21 soil bacterial isolates in equal proportions) and high-diversity natural top-soil communities followed highly reproducible succession paths, maintaining distinct soil microbiome signatures. Development trajectories and compositional states were different for communities propagated in soils than in liquid suspension. Microbiome states were maintained over multiple renewed growth cycles but could be diverged by short-term pollutant exposure. The different but robust trajectories demonstrated that deterministic taxa-inherent characteristics underlie reproducible development and self-organized complexity of soil microbiomes within their environmental boundary conditions. Our findings also have direct implications for potential strategies to achieve controlled restoration of desertified land. TEASERSpecies-rich soil microbiomes grow and propagate reproducibly despite inherent stochastic complexity, paving the way for soil restoration.

microbiology↗

bioRxiv2018

Rapid inference of direct interactions in large-scale ecological networks from heterogeneous microbial sequencing data

The recent explosion of metagenomic sequencing data opens the door towards the modeling of microbial ecosystems in unprecedented detail. In particular, co-occurrence based prediction of ecological interactions could strongly benefit from this development. However, current methods fall short on several fronts: univariate tools do not distinguish between direct and indirect interactions, resulting in excessive false positives, while approaches with better resolution are so far computationally highly limited. Furthermore, confounding variables typical for cross-study data sets are rarely addressed. We present FlashWeave, a new approach based on a flexible Probabilistic Graphical Models framework to infer highly resolved direct microbial interactions from massive heterogeneous microbial abundance data sets with seamless integration of metadata. On a variety of benchmarks, FlashWeave outperforms state-of-the-art methods by several orders of magnitude in terms of speed while generally providing increased accuracy. We apply FlashWeave to a cross-study data set of 69 818 publicly available human gut samples, resulting in one of the largest and most diverse models of microbial interactions in the human gut to date.

systems biology↗

bioRxiv2017

MAPseq: Improved Speed, Accuracy And Consistency In Ribosomal RNA Sequence Analysis

Metagenomic sequencing has become crucial to studying microbial communities, but meaningful taxonomic analysis and inter-comparison of such data are still hampered by technical limitations, between-study design variability and inconsistencies between taxonomies used. Here we present MAPseq, a framework for reference-based rRNA metagenomic analysis that is up to 30% more accurate (F1/2 score) and up to one hundred times faster than existing solutions, providing in a single run multiple taxonomy classifications and hierarchical OTU mappings, for both amplicon and shotgun sequencing strategies, and for datasets of virtually any size. Availability: Source code and binaries are freely available at http://meringlab.org/software/mapseq/

bioinformatics↗