Biology subjects

Cirolia, G.

Publications and source records attributed to Cirolia, G..

3 recordsLinked to original sources

bioRxiv2026

Performance of IBD machine learning classifiers varies across microbiome training data independent of geographic diversity

Microbiome-based machine learning classifiers show increasing promise for disease identification across gastrointestinal, metabolic, and immune-mediated conditions. Inflammatory bowel disease (IBD), a chronic immune-mediated disorder associated with disruption of the gut microbiome, has been a particularly successful application area. However, while many predictive models achieve high performance within individual datasets, their ability to generalize across independent populations and geographic contexts remains unclear. Here, we tested whether model class and training dataset composition influence model generalizability across geographically diverse evaluation studies. We compiled seven publicly available shotgun metagenomic studies spanning five geographic regions, comprising 697 individuals with IBD or healthy controls. We trained 246,986 model configurations across seven model classes and five distinct training dataset combinations and evaluated top-performing models on independent studies from the USA, Ireland, Germany, Israel and China. Extreme gradient boosting and random forest models showed the highest and most consistent performance across training datasets, a ranking that was maintained on independent evaluation studies. However, models trained on geographically diverse datasets did not outperform those trained on USA-only datasets. Instead, model performance was strongly dependent on the evaluation study itself, with consistent differences in achievable accuracy across studies. Despite most models achieving similar AUC scores, there was limited overlap in the key microbial species identified. Furthermore, even for the small set of disease predictive microbes shared between models, the direction of enrichment between IBD or healthy subjects often varied in opposing directions across study populations. These findings suggest that study-specific factors constrain generalization and may help explain the lack of consistent microbiome-based biomarkers for IBD.

microbiology↗

bioRxiv2020

CRISPR-based functional genomics in human dendritic cells

Dendritic cells (DCs) regulate processes ranging from antitumor and antiviral immunity to host-microbe communication at mucosal surfaces. It remains difficult, however, to genetically manipulate human DCs, limiting our ability to probe how DCs elicit specific immune responses. Here, we develop a CRISPR/Cas9 genome editing method for human monocyte-derived DCs (moDCs) that mediates knockouts with a median efficiency of >93% across >300 genes. Using this method, we perform genetic screens in moDCs, identifying mechanisms by which DCs tune responses to lipopolysaccharides from the human microbiome. In addition, we reveal donor-specific responses to lipopolysaccharides, underscoring the importance of assessing immune phenotypes in donor-derived cells, and identify genes that control this specificity, highlighting the potential of our method to pinpoint determinants of inter-individual variation in immune responses. Our work sets the stage for a systematic dissection of the immune signaling at the host-microbiome interface and for targeted engineering of DCs for neoantigen vaccination.

immunology↗

bioRxiv2019

Methods Matter -- Standard Production Platforms For Recombinant AAV Can Produce Chemically And Functionally Distinct Vectors

Different manufacturing approaches have been used in the production of recombinant adeno-associated virus (rAAV). The two leading approaches are transiently transfected human HEK293 cells and live baculovirus infection of Sf9 insect cells. Unexplained differences in vector performance have been seen clinically and preclinically. Thus, we performed for the first time a highly controlled comparative production analysis varying only the host cell species but keeping all other rAAV production parameters the same. We demonstrate that host cell species is critical for determining vector potency. Given these key findings, we then sought to deeply characterize differences in rAAVs when produced by these two manufacturing platforms with multiple analytical approaches including: proteomic profiling by mass spectrometry, isoelectric focusing, cryo-EM, denaturation assays, genomic and epigenomic sequencing of packaged genomes, human cytokine profiling, and comparative functional transduction assessments in vitro and in vivo, including in humanized liver mice. Using these tools weve made two major discoveries: 1) rAAV capsids have post-translational modifications (PTMs) including glycosylation, acetylation, phosphorylation, methylation and deamidation, and these PTMs differ between platforms; 2) rAAV genomes are methylated during production, and these methylation marks are also differentially deposited between platforms. In addition, our data also demonstrate that host cell protein impurities differ between platforms and can have their own PTMs including potentially immunogenic N-linked glycans. We show that human-produced rAAVs are more potent than baculovirus-Sf9 vectors in various cell types in vitro (P < 0.05-0.0001), in various mouse tissues in vivo (P < 0.03-0.0001), and in human liver in vivo (P < 0.005). Collectively, our findings were reproducible across vendors, including commercial manufacturers, academic core facilities, and individual laboratory preparations. These vector differences may have clinical implications for rAAV receptor binding, trafficking, expression kinetics, expression durability, vector immunogenicity as well as cost considerations.

molecular biology↗