Biology subjects

Lorenz, P.

Publications and source records attributed to Lorenz, P..

6 recordsLinked to original sources

bioRxiv2025

Breaking Through Biology's Data Wall: Expanding the Known Tree of Life by Over 10x using a Global Biodiscovery Pipeline

Advancements in the life sciences have always been built upon our collective understanding of life on Earth. Now, the rise of generative biology - the use of AI foundation models to design, generate, and annotate proteins, pathways and therapeutics - is creating unprecedented demand for large, diverse biological sequence datasets. While a limited subset of such data can be generated in clinical or laboratory settings, the vast majority of the training data for unsupervised models must be sourced from the natural world - the product of nearly four billion years of evolutionary history. However, the public databases that currently supply this data, while foundational to research, were established to aggregate results from academic experiments, not as training datasets for machine learning. Their human-centric data structure limits model performance due to redundancy, taxonomic and geographic bias, limited biological context, and inconsistent provenance. With 68% of all sequence data in the SRA database coming from just 5 species, this is one of the most severe class imbalance problems ever encountered in AI. Legal and infrastructural constraints further exacerbate this bottleneck. To address these limitations and support scalable model training, we introduce BaseData: the largest and fastest-growing biological sequence database ever built, and the first purpose-built for training foundation models. As of late 2024, BaseData contained 9.8 billion novel genes, representing more than a 10-fold expansion in known protein diversity after accounting for redundancy. BaseData also contains more than 1 million species not represented in other genomic databases. Its partnership-driven data supply chain across 26 countries and autonomous regions enables growth of up to 2 billion novel genes per month, far exceeding public repositories. All data is collected under benefit sharing agreements using standardized protocols and structured using graph-based, ontology-rich metadata that preserves evolutionary context. BaseData represents a new, ethically-grounded infrastructure for training biological foundation models, complementing public efforts and enabling the next era of generative biology. Short AbstractProgress of AI in biology is now being limited by the availability of high-quality biological sequence data from nature. To get past this data wall, we introduce BaseData, a new biological database built on top of a global, scalable biodiscovery pipeline. As of 2024, BaseData had already expanded the known protein universe by over 10x.

genomics↗

bioRxiv2024

Conditional language models enable the efficient design of proficient enzymes

The design of functional enzymes holds promise for transformative solutions across various domains but presents significant challenges. Inspired by the success of language models in generating nature-like proteins, we explored the potential of an enzyme-specific language model in designing catalytically active artificial enzymes. Here, we introduce ZymCTRL (enzyme control), a conditional language model trained on the enzyme sequence space, capable of generating enzymes based on user-defined specifications. Experimental validation at diverse data regimes and for different enzyme families demonstrated ZymCTRLs ability to generate active enzymes across various sequence identity ranges. Specifically, we describe the design of carbonic anhydrases and lactate dehydrogenases in zero-shot, without requiring further training of the model, and showcasing activity at sequence identities below 40% compared to natural proteins. Biophysical analysis confirmed the globularity and well-folded nature of the generated sequences. Furthermore, fine-tuning the model enabled the generation of lactate dehydrogenases outside of natural sequence space but with activity comparable to their natural counterparts. Two of the artificial lactate dehydrogenases were selected for scale production and successfully lyophilised, maintaining activity and demonstrating preliminary conversion in one-pot enzymatic cascades under extreme conditions. Our findings open a new door towards the rapid and cost-effective design of artificial proficient enzymes. The model and dataset are freely available to the community.

bioinformatics↗

bioRxiv2024

IMPROVING ALPHAFOLD2 PERFORMANCE WITH A GLOBAL METAGENOMIC & BIOLOGICAL DATA SUPPLY CHAIN

Scaling laws suggest that more than a trillion species inhabit our planet but only a miniscule and unrepresentative fraction (less than 0.00001%) have been studied or sequenced to date. Deep learning models, including those applied to tasks in the life sciences, depend on the quality and size of training or reference datasets. Given the large knowledge gap we experience when it comes to life on earth, we present a data-centric approach to improving deep learning models in Biology: We built partnerships with nature parks and biodiversity stakeholders across 5 continents covering 50% of global biomes, establishing a global metagenomics and biological data supply chain. With higher protein sequence diversity captured in this dataset compared to existing public data, we apply this data advantage to the protein folding problem by MSA supplementation during inference of AlphaFold2. Our model, BaseFold, exceeds traditional AlphaFold2 performance across targets from the CASP15 and CAMEO, 60% of which show improved pLDDT scores and RMSD values being reduced by up to 80%. On top of this, the improved quality of the predicted structures can yield better docking results. By sharing benefits with the stakeholders this data originates from, we present a way of simultaneously improving deep learning models for biology and incentivising protection of our planets biodiversity.

genomics↗

bioRxiv2023

Piezo inhibition prevents and rescues scarring by targeting the adipocyte to fibroblast transition

While past studies have suggested that plasticity exists between dermal fibroblasts and adipocytes, it remains unknown whether fat actively contributes to fibrosis in scarring. We show that adipocytes convert to scar-forming fibroblasts in response to Piezo-mediated mechanosensing to drive wound fibrosis. We establish that mechanics alone are sufficient to drive adipocyte-to- fibroblast conversion. By leveraging clonal-lineage-tracing in combination with scRNA-seq, Visium, and CODEX, we define a "mechanically naive" fibroblast-subpopulation that represents a transcriptionally intermediate state between adipocytes and scar-fibroblasts. Finally, we show that Piezo1 or Piezo2-inhibition yields regenerative healing by preventing adipocytes activation to fibroblasts, in both mouse-wounds and a novel human-xenograft-wound model. Importantly, Piezo1-inhibition induced wound regeneration even in pre-existing established scars, a finding that suggests a role for adipocyte-to-fibroblast transition in wound remodeling, the least-understood phase of wound healing. Adipocyte-to-fibroblast transition may thus represent a therapeutic target for minimizing fibrosis via Piezo-inhibition in organs where fat contributes to fibrosis.

cell biology↗

bioRxiv2021

Mapping Human Transient Transcriptomes Using Single Nucleotide Resolution 4sU Sequencing (SNU-Seq)

Genomes are pervasively transcribed, leading to stable and unstable transcripts that influence 3-dimensional genome organisation and gene regulation. High sensitivity and nucleotide resolution are required to resolve mammalian nascent transcriptomes. Here, we exploit the sensitivity of 4-thio-uridine (4sU) metabolic pulse-labelling to develop two nucleotide-resolution methods: Single-Nucleotide resolution 4sU sequencing (SNU-Seq) and size-fractionated 4sU-Seq (sf4sU-Seq). sf4sU-Seq involves gel isolation of abundant 4sU-labelled promoter proximal nascent transcripts, enabling nucleotide resolution mapping of transcription start sites and promoter proximal pauses (PPPs) on the same transcript. SNU-Seq exploits 3 end RNA-Seq, using bacterial poly(A) polymerase (bPAP) to polyadenylate the 3 ends of nascent transcripts and create oligo(dT)-primed libraries. The artificial poly(A) tail marks the precise position of polymerase on a transcription unit. SNU-Seq read levels are similar at pre-mRNAs and enhancers genome-wide and read spikes in pre-mRNA outputs map pauses, PPPs and polyadenylation sites. SNU-Seq enables discovery of thousands of unannotated regions of divergent transcription and helps define hundreds of the more than 10,000 regions of primed non-transcribed acetylated open chromatin that induce divergent nascent transcripts within 0.5h of IFN-{gamma} treatment in Hep3B cells. Thus, combining chromatin analysis with SNU-Seq reveals the transcriptional responsiveness of an epigenetically primed human genome. HIGHLIGHTSO_LISNU-Seq maps nascent transcripts with bp resolution, high sensitivity and low cost C_LIO_LIsf4sU-Seq resolves TSS and PPP at the same gene, complementing SNU-Seq C_LIO_LI1000s of divergently transcribed enhancers resolved by SNU-Seq C_LIO_LIRapid IFN{gamma} dependent transcriptional induction from primed Hep3B epigenome C_LI GRAPHICAL ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=165 SRC="FIGDIR/small/452379v2_ufig1.gif" ALT="Figure 1"> View larger version (47K): org.highwire.dtl.DTLVardef@da6239org.highwire.dtl.DTLVardef@16692c6org.highwire.dtl.DTLVardef@12e30f7org.highwire.dtl.DTLVardef@190c98f_HPS_FORMAT_FIGEXP M_FIG C_FIG

molecular biology↗

bioRxiv2019

H3K4me3 is neither instructive for, nor informed by, transcription.

H3K4me3 is a near-universal histone modification found predominantly at the 5 region of genes, with a well-documented association with gene activity. H3K4me3 has been ascribed roles as both an instructor of gene expression and also a downstream consequence of expression, yet neither has been convincingly proven on a genome-wide scale. Here we test these relationships using a combination of bioinformatics, modelling and experimental data from budding yeast in which the levels of H3K4me3 have been massively ablated. We find that loss of H3K4me3 has no effect on the levels of nascent transcription or transcript in the population. Moreover, we observe no change in the rates of transcription initiation, elongation, mRNA export or turnover, or in protein levels, or cell-to-cell variation of mRNA. Loss of H3K4me3 also has no effect on the large changes in gene expression patterns that follow galactose induction. Conversely, loss of RNA polymerase from the nucleus has no effect on the pattern of H3K4me3 deposition and little effect on its levels, despite much larger changes to other chromatin features. Furthermore, large genome-wide changes in transcription, both in response to environmental stress and during metabolic cycling, are not accompanied by corresponding changes in H3K4me3. Thus, despite the correlation between H3K4me3 and gene activity, neither appear to be necessary to maintain levels of the other, nor to influence their changes in response to environmental stimuli. When we compare gene classes with very different levels of H3K4me3 but highly similar transcription levels we find that H3K4me3-marked genes are those whose expression is unresponsive to environmental changes, and that their histones are less acetylated and dynamically turned-over. Constitutive genes are generally well-expressed, which may alone explain the correlation between H3K4me3 and gene expression, while the biological role of H3K4me3 may have more to do with this distinction in gene class.

genomics↗