Biology subjects

Jensen, P. A.

Publications and source records attributed to Jensen, P. A..

10 recordsLinked to original sources

bioRxiv2025

Planning and scheduling biological experiments across multiple liquid handling robots

Coordinating multiple liquid handling robots is a complex logistical task when designing biological experiments. Protocol designers must consider the capabilities and constraints of each robot to distribute work optimally across multiple instruments. We developed an optimization framework that finds optimal liquid handling solutions that leverage an arbitrary number of robots. Our algorithm, called Pourfecto, abstracts the capabilities of each robot and their labware, allowing us to plan and schedule a wide range of biological experiments using commercial instruments and custom-built hardware. Pourfecto can optimize multiple objectives (minimum transfers, fewest reagents, fewest labware swaps) and scales to experiments with hundreds of thousands of liquid transfers.

synthetic biology↗

bioRxiv2025

Monotherapy cancer drug-blind response prediction is limited to intraclass generalization

Monotherapy cancer drug response prediction (DRP) models predict the response of a cell line to a given drug. Analyzing these models performance includes assessing their ability to predict the response of cell lines to new drugs, i.e. drugs that are not in the training set. Drug-blind prediction displays greatly diminished performance or outright failure across a wide range of model architectures and different large pharmacogenomic datasets. Drug-blind failure is hypothesized to be caused by the relatively limited set of drugs present in these datasets. The time and cost associated with further cell line experiments is significant, and it is impossible to predict beforehand how much data would be enough to overcome drug-blind failure. We must first define how current data contributes to drug-blind failure before attempting to remedy drug-blind failure with further data collection. In this work, we quantify the extent to which drug-blind generalizability relies on mechanistic overlap of drugs between training and testing splits. We first identify that the majority of mixed set DRP model performance can be attributed to drug overfitting, likely inhibiting generalization and preventing accurate analysis. Then, by specifically probing the drug-blind ability of models, we reveal the sources of generalizable drug features are confined to shared mechanisms of action and related pathways. Furthermore, we observed that, for certain mechanisms, we can significantly improve performance by limiting the training of models to a single mechanism compared to training on all drugs simultaneously. We conclude that drug-blind performance is a poor benchmark for DRP as it does not describe model behavior, it describes dataset behavior. Our investigation displays that deep learning models trained on large, monotherapy cell line panels can more accurately describe mechanism of action of drugs rather than their advertised connection to broader cancer biology. Author summaryIn this paper, we characterize the feature space of cancer drug-blind prediction. To understand the efficacy of a novel cancer drug it has never seen before (drug-blind), a model must be able to accurately compare this drug to all drugs it saw during training. These relationships between cancer drugs, the feature space, must be described well enough that this is possible. We believe that these relationships are poorly defined because cancer DRP models always display reduced performance in a drug-blind context. For the first time, we quantified the limits of generalization in a drug-blind setting. We showed that drug-blind generalization describes mechanistic relationships among drugs during model training. We also outlined new criteria with which to judge the drug-blind ability of a model. Failure of drug-blind prediction is an oft overlooked shortcoming in cancer DRP with potentially damaging downstream implications. We hope to show drug-blind ability of these models in a new light to guide others towards more pertinent tasks in cancer deep learning.

bioinformatics↗

bioRxiv2025

Ten species comprise half of the bacteriology literature, leaving most species unstudied

Microbiology research has historically focused on a few species of model organisms. Our bibliographic analysis finds extreme bias in the distribution of bacteriology research across species, with half of all papers referencing only ten species and 74% of all known species remaining unstudied. Microbiologists will need to broaden their perspective and embrace complexity to develop a complete understanding of the microbial world.

microbiology↗

bioRxiv2024

The MutRS quorum sensing system controls lantibiotic mutacin production in the human pathogen Streptococcus mutans

Microbes use quorum sensing systems to respond to ecological and environmental changes. In the oral microbiome, the pathogenic bacterium Streptococcus mutans uses quorum sensing to control the production of bacteriocins. These antimicrobial peptides kill off ecological competetors and allow S. mutans to dominate the microenviornment of dental plaques and form dental caries. One class of bacteriocins produced by S. mutans, the lantibiotic mutacins, are particularly effective at killing due to their broad spectrum of activity. Despite years of study, the regulatory mechanisms governing production of lantibiotic mutacins I, II, and III in S. mutans have never been elucidated. We discovered a distinct class of quorum sensing system, MutRS, that regulates mutacins and is widespread among the streptococci. We demonstrate that MutRS systems are activated by a short peptide pheromone (Mutacin Stimulating Peptide, or MSP) and show that MutRS controls production of three separate lantibiotic mutacins in three different strains of S. mutans. Finally, we show that paralogous MutRS systems participate in inter- and intrastrain crosstalk, providing further evidence of the interplay between quorum sensing systems in the oral streptococci.

microbiology↗

bioRxiv2024

The pneumococcal bacteriocin streptococcin B is produced as part of the early competence cascade and promotes intraspecies competition.

Streptococcus pneumoniae is an important human pathogen that normally resides in the human nasopharynx. Competence mediated bacteriocin expression by S pneumoniae plays a major role in both the establishment and persistence of colonization on this polymicrobial surface. Over 20 distinct bacteriocin loci have been identified in pneumococcal genomes, but only a small number have been characterized phenotypically. In this work, we demonstrate that 3/4th of S. pneumoniae strains contain a highly conserved scb locus that encodes an active lactococcin 972-like bacteriocin called Streptococcin B. In these backgrounds, the scbABC locus is part of the early competence cascade due to a ComE binding site in the promoter region. Streptococcin B producing strains target both members of the population that have failed to activate competence and the 25% of the population that carry a naturally occurring deletion of the ComE binding site and the functional bacteriocin gene. The ComR-type regulator found directly upstream of the scb locus in S. pneumoniae strains can activate scb expression independent of the presence of the ComE binding site but only when stimulated by a peptide that is encoded in the scb locus of S. pseudopneumoniae, a closely related bacterium which also inhabits the human nasopharynx. Given the co-regulation with competence and the phenotypic confirmation of activity, Streptococcin B represents a previously unrecognized fratricide effector that gives producing strains an additional advantage over the naturally occurring deleted strains during colonization. ImportanceStreptococcus pneumoniae is a common cause of pneumonia, meningitis, sinusitis and otitis media. In order to successfully colonize humans, a pre-requisite to the development of invasive disease, S. pneumoniae must compete with other bacterial inhabitants of the nasal surface for space and nutrients. Bacteriocins are small antimicrobial peptides produced by bacteria that typically target neighboring bacteria by disruption of the cell surface. S. pnuemoniae encodes a large number of potential bacteriocin, but, for most, their role in competitive interactions has not been defined. This work demonstrates that isolates that produce the bacteriocin Streptococcin B have an advantage over non producers. These observations contribute to our understanding of the competitive interactions that precede development of S. pneumoniae disease.

microbiology↗

bioRxiv2021

A Reinforcement Learning Framework for Pooled Oligonucleotide Design

The goal of oligonucleotide (oligo) design is to select oligos that optimize a set of design criteria. Oligo design problems are combinatorial in nature and require computationally intensive models to evaluate design criteria. Even relatively small problems can be intractable for brute-force approaches that test every possible combination of oligos, so heuristic approaches must be used to find near-optimal solutions. We present a general reinforcement learning framework, called OligoRL, to solve oligo design problems with complex constraints. OligoRL allows "black-box" design criteria and can be adapted to solve many oligo design problems. We highlight the flexibility of OligoRL by building tools to solve three distinct design problems: 1.) finding pools of random DNA barcodes that lack restriction enzyme recognition sequences (CutFreeRL); 2.) compressing large, non-degenerate oligo pools into smaller degenerate ones (OligoCompressor); and 3.) finding Not-So-Random hexamer primer pools that avoid rRNA and other unwanted transcripts during RNA-seq library preparation (NSR-RL). OligoRL demonstrates how reinforcement learning offers a general solution for complex oligo design problems. OligoRL and its associated software tools are available as a Julia package at http://jensenlab.net/tools.

bioinformatics↗

bioRxiv2020

A ComRS competence pathway in the oral pathogen Streptococcus sobrinus

Streptococcus sobrinus is one of two species of bacteria that cause dental caries (tooth decay) in humans. Our knowledge of S. sobrinus is limited despite the organisms important role in oral health. It is widely believed that S. sobrinus lacks the natural competence pathways that are used by other streptococci to regulate growth, virulence, and quorum sensing. The lack of natural competence has also prevented genetic manipulation of S. sobrinus, limiting our knowledge of its pathogenicity. We discovered a functional ComRS competence system in S. sobrinus. The ComRS pathway in S. sobrinus has a unique structure, including two copies of the transcriptional regulator ComR and a peptide pheromone (XIP) that lacks aromatic amino acids. We show that synthetic XIP allows transformation of S. sobrinus with plasmid or linear DNA, and we leverage this newfound genetic tractability to confirm that only one of the ComR homologs is required for induced competence. Although S. sobrinus is typically placed among the mutans group streptococci, the S. sobrinus ComRS system is structurally and functionally similar to the competence pathways in the salivarius group. Like S. salivarius, the ComRS gene cluster in S. sobrinus includes a peptide cleavage/export gene, and the ComRS system appears coupled to a bacteriocin response system. These findings raise questions about the true phylogenetic placement of S. sobrinus. Finally, we identified two strains of S. sobrinus appear to be "cheaters" by either not responding to or not producing XIP. While the mechanisms of cheating could be independent, we show how a recombination event in the non-responsive strain would restore function of the ComRS pathway but delete the gene encoding XIP. Thus the S. sobrinus ComRS pathway provides a lens into the evolution of ecological cheaters.

microbiology↗

bioRxiv2019

gapsplit: Efficient random sampling for non-convex constraint-based models

SummaryO_SCPCAPGC_SCPCAPO_SCPLOWAPSPLITC_SCPLOW generates random samples from convex and non-convex constraint-based models. O_SCPCAPGC_SCPCAPO_SCPLOWAPSPLITC_SCPLOW targets under-sampled regions of the solution space for uniform coverage.\n\nAvailability and ImplementationPython and Matlab source code are freely available at http://jensenlab.net/tools.\n\nContactpjens@illinois.edu

systems biology↗

bioRxiv2019

Efficient enzyme coupling algorithms identify functional pathways in genome-scale metabolic models.

Flux coupling identifies sets of reactions whose fluxes are \"coupled\" or correlated in genome-scale models. By identified sets of coupled reactions, modelers can 1.) reduce the dimensionality of genome-scale models, 2.) identify reactions that must be modulated together during metabolic engineering, and 3.) identify sets of important enzymes using high-throughput data. We present three computational tools to improve the efficiency, applicability, and biological interpretability of flux coupling analysis.\n\nThe first algorithm (cachedFCF) uses information from intermediate solutions to decrease the runtime of standard flux coupling methods by 10-100 fold. Importantly, cachedFCF makes no assumptions regarding the structure of the underlying model, allowing efficient flux coupling analysis of models with non-convex constraints.\n\nWe next developed a mathematical framework (FALCON) that incorporates enzyme activity as continuous variables in genome-scale models. Using data from gene expression and fitness assays, we verified that enzyme sets calculated directly from FALCON models are more functionally coherent than sets of enzymes collected from coupled reaction sets.\n\nFinally, we present a method (delete-and-couple) for expanding enzyme sets to allow redundancies and branches in the associated metabolic pathways. The expanded enzyme sets align with known biological pathways and retain functional coherence. The expanded enzyme sets allow pathway-level analyses of genome-scale metabolic models.\n\nTogether, our algorithms extend flux coupling techniques to enzymatic networks and models with transcriptional regulation and other non-convex constraints. By expanding the efficiency and flexibility of flux coupling, we believe this popular technique will find new applications in metabolic engineering, microbial pathogenesis, and other fields that leverage network modeling.

systems biology↗

bioRxiv2018

Metabolic modeling of Streptococcus mutans reveals complex nutrient requirements of an oral pathogen

Streptococcus mutans is a Gram positive bacterium that thrives under acidic conditions and is a primary cause of tooth decay (dental caries). To better understand the metabolism of S. mutans on a systematic level, we manually constructed a genome-scale metabolic model of the S. mutans type strain UA159. The model, called iSMU, contains 656 reactions involving 514 metabolites and the products of 488 genes.\n\nWe interrogated S. mutans nutrient requirements using model simulations and nutrient removal experiments in defined media. The iSMU model matched experimental results in greater than 90% of the conditions tested. We also simulated effects of single gene deletions. The models predictions agreed with 78.1% and 84.4% of the gene essentiality predictions from two experimental datasets. Our manually curated model is more accurate than S. mutans models generated from automated reconstruction pipelines. We believe the iSMU model is an important resource for understanding how metabolism enables the cariogenicity of S. mutans.

systems biology↗