Johan Pensar

• Cibinel, Luca; Linjordet, Trond; Pensar, Johan; Badia, David Balcells; Bin, Riccardo De & Ell, Basil (2026). tmQM-RDF Data Set: A Knowledge Graph Representing Transition Metal Complexes. Journal of Chemical Information and Modeling. ISSN 1549-9596. 66(13), p. 7524–7538. doi: 10.1021/acs.jcim.6c01281. Full text in Research Archive Show summary

  Transition metal complexes (TMCs) have wide-ranging practical utility in chemistry, with possible applications that range from catalysis to medicinal chemistry. The study of TMCs and their properties is thus a field rich with potential, one in which machine learning and computational approaches can offer a substantial aid. For this reason, appropriate and accessible data sets, collecting a wide range of information, are required in order to facilitate the effective analysis and investigation of such compounds. This paper contributes to the data modeling effort via the introduction of the transition metal quantum mechanics RDF (tmQM-RDF) data set, a knowledge graph constructed using the Resource Description Framework (RDF) vocabulary which collects rich and detailed descriptions of approximately 60k TMCs. These descriptions are both qualitative and quantitative, encompassing the compositional nature of TMCs as well as the entirety of their molecular graphs. The Python package tmqmrdfdata is released together with the knowledge graph to provide a user-friendly interface. An example of the power of the proposed representation is presented, showcasing how the information available in tmQM-RDF can be exploited for TMC manipulation tasks, achieving promising performance even with relatively simple probabilistic models.

• Meggio, Claudio; Pensar, Johan; Badia, David Balcells & Bin, Riccardo De (2026). A path-based boosting algorithm for exploring transition metal compounds. Statistical Modelling. ISSN 1471-082X. doi: 10.1177/1471082x251407458. Full text in Research Archive Show summary

  Motivated by the analysis of chemical metal compounds and their properties for catalysis, we developed a gradient boosting model that explores graph structures to perform prediction tasks. Taking advantage of the iterative nature of boosting, our novel approach, called PathBoost , explores the graphs to identify the relevant paths and simultaneously fits a prediction model. Advantages of PathBoost include automatic variable selection, as only relevant paths are kept in the model and explainability, as a measure of variable importance is provided. The novel algorithm is applied to the tmQM dataset, where the molecules are represented as graphs with atoms as nodes and bonds as edges. The goal is to predict specific quantum properties of the molecules, such as the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) gap. These properties usually require heavy computational power to be computed, while our model aims to provide comparable results using much fewer resources.

• Pérez, Johan Manuel de Aguas; Henckel, Leonard; Pensar, Johan & Biele, Guido (2025). Causal Inference amid Missingness-Specific Independencies and Mechanism Shifts. Proceedings of Machine Learning Research (PMLR). 286, p. 31–42. doi: https:/proceedings.mlr.press/v286/aguas25a.html. Full text in Research Archive Show summary

  The recovery of causal effects in structural models with missing data often relies on m-graphs, which assume that missingness mechanisms do not directly influence substantive variables. Yet, in many real-world settings, missing data can alter decision-making processes, as the absence of key information may affect downstream actions and states. To overcome this limitation, we introduce lm-SCMs and lm-graphs, which extend m-graphs by integrating a label set that represents relevant context-specific independencies (CSI), accounting for mechanism shifts induced by missingness. We define two causal effects within these systems: the Full Average Treatment Effect (FATE), which reflects the effect in a hypothetical scenario had no data been missing, and the Natural Average Treatment Effect (NATE), which captures the effect under the unaltered CSIs in the system. We propose recovery criteria for these queries and present doubly-robust estimators for a graphical model inspired by a real-world application. Simulations highlight key differences between these estimands and estimation methods. Findings from the application case suggest a small effect of ADHD treatment upon test achievement among Norwegian children, with a slight effect shift due to missing pre-tests scores.

• Kuronen, Juri; Corander, Jukka & Pensar, Johan (2025). Learning pairwise Markov network structures using correlation neighborhoods. Communications in Statistics - Simulation and Computation. ISSN 0361-0918. doi: 10.1080/03610918.2025.2602032. Full text in Research Archive Show summary

  Markov networks are widely studied and used throughout multivariate statistics and computer science. In particular, the problem of learning the structure of Markov networks from data without invoking chordality assumptions in order to retain expressiveness of the model class has been given a considerable attention in the recent literature, where numerous constraint-based or score-based methods have been introduced. Here we develop a new search algorithm for the network score-optimization that has several computational advantages and scales well to high-dimensional data sets. The key observation behind the algorithm is that the neighborhood of a variable can be efficiently captured using local penalized likelihood ratio (PLR) tests by exploiting an exponential decay of correlations across the neighborhood with an increasing graph-theoretic distance from the focus node. The candidate neighborhoods are then processed by a two-stage hill-climbing (HC) algorithm. Our approach, termed fully as PLRHC-BIC0.5, compares favorably against the state-of-the-art methods in all our experiments spanning both low- and high-dimensional networks and a wide range of sample sizes. An efficient implementation of PLRHC-BIC0.5 is freely available from the URL: https://github.com/jurikuronen/plrhc.

• Kvisgaard, Vera Haugen & Pensar, Johan (2025). Bayesian Estimation of Causal Effects from Observational Categorical Data. Bayesian Analysis. ISSN 1936-0975. p. 1–28. doi: 10.1214/25-BA1540. Full text in Research Archive Show summary

  We present a Bayesian procedure for estimation of pairwise intervention effects in a high-dimensional system of categorical variables. We assume that we have observational data generated from an unknown causal Bayesian network for which there are no latent confounders. Most of the existing methods developed for this setting assume that the underlying model is linear Gaussian, including the Bayesian IDA (BIDA) method that we build upon in this work. By combining a Bayesian backdoor estimator with model averaging, we obtain a posterior over the intervention distributions of a cause-effect pair that can be expressed as a mixture over stochastic linear combinations of Dirichlet distributions. Although there is no tractable closed-form expression for the posterior density, it is straightforward to produce Monte Carlo approximations of target quantities through direct sampling, and we also derive closed-form expressions for a few selected moments. To scale up the proposed procedure, we employ Markov Chain Monte Carlo (MCMC), which also enables us to use more efficient adjustment sets compared to the current exact BIDA. Finally, we use Jensen-Shannon divergence to define a novel causal effect based on a set of intervention distributions in the general categorical setting. We compare our method to the original IDA method and existing Bayesian approaches in numerical simulations and show that BIDA performs favorably against these alternative methods in terms of producing point estimates and discovering strong effects.

• Pérez, Johan Manuel de Aguas; Henckel, Leonard; Pensar, Johan & Biele, Guido (2025). Causal Inference amid Missingness-Specific Independences and Mechanism Shifts. Proceedings of Machine Learning Research (PMLR). 286, p. 31–44. Full text in Research Archive
• Pérez, Johan Manuel de Aguas; Krumscheid, Sebastian; Pensar, Johan & Biele, Guido (2025). The Probability of Tiered Benefit: Partial Identification with Robust and Stable Inference. Proceedings of Machine Learning Research (PMLR). 275, p. 90–113. doi: https:/proceedings.mlr.press/v275/aguas25a.html. Full text in Research Archive
• Kvisgaard, Vera Haugen & Pensar, Johan (2025). Bayesian IDA with Background Knowledge. Italian Statistical Society Series on Advances in Statistics. ISSN 3059-2135. p. 91–96. doi: 10.1007/978-3-031-96736-8_16. Full text in Research Archive Show summary

  There have been several recent advancements in the development of causal learning methods that combine structure learning and causal inference. Inspired by the frequentist IDA method, the Bayesian IDA (BIDA) method is a scalable Bayesian approach for estimating causal effects from observational data under an unknown causal Bayesian network. Here, we extend BIDA to the setting where the causal structure is partly known.

• Hjort, Anders Dahl; Hermansen, Gudmund Horn; Pensar, Johan & Williams, Jonathan P. (2025). Uncertainty quantification in automated valuation models with spatially weighted conformal prediction. International Journal of Data Science and Analytics (JDSA). ISSN 2364-415X. 20, p. 7089–7106. doi: 10.1007/s41060-025-00862-4. Full text in Research Archive Show summary

  Nonparametric machine learning models, such as random forests and gradient boosted trees, are frequently used to estimate house prices due to their predictive accuracy, but a main drawback of such methods is their limited ability to quantify prediction uncertainty. Conformal prediction (CP) is a model-agnostic framework for constructing confidence sets around predictions of machine learning models with minimal assumptions. However, due to the spatial dependencies observed in house prices, direct application of CP leads to confidence sets that are not calibrated everywhere, i.e., the confidence sets will be too large in certain geographical regions and too small in others. We survey various approaches to adjust the CP confidence set to account for this and demonstrate their performance on a data set from the housing market in Oslo, Norway. Our findings indicate that calibrating the confidence sets on a spatially weighted version of the non-conformity scores makes the coverage more consistently calibrated across geographical regions. We also perform a simulation study on synthetically generated sale prices to empirically explore the performance of CP on housing market data under idealized conditions with known data-generating mechanisms.

• Pérez, Johan Manuel de Aguas; Pensar, Johan; Pérez, Tomás Fernando Varnet & Biele, Guido (2025). Recovery and inference of causal effects with sequential adjustment for confounding and attrition. Journal of Causal Inference. ISSN 2193-3677. 13(1). doi: 10.1515/jci-2024-0009. Full text in Research Archive
• Hajj, Ghadi al; Hubin, Aliaksandr; Kanduri, Chakravarthi; Pavlovic, Milena; Rand, Knut Dagestad & Widrich, Michael [Show all 11 contributors for this article] (2024). Incorporating probabilistic domain knowledge into deep multiple instance learning. Proceedings of Machine Learning Research (PMLR). 235, p. 17279–17297. Full text in Research Archive Show summary

  Deep learning methods, including deep multiple instance learning methods, have been criticized for their limited ability to incorporate domain knowledge. A reason that knowledge incorporation is challenging in deep learning is that the models usually lack a mapping between their model components and the entities of the domain, making it a non-trivial task to incorporate probabilistic prior information. In this work, we show that such a mapping between domain entities and model components can be defined for a multiple instance learning setting and propose a framework DeeMILIP that encompasses multiple strategies to exploit this mapping for prior knowledge incorporation. We motivate and formalize these strategies from a probabilistic perspective. Experiments on an immune-based diagnostics case show that our proposed strategies allow to learn generalizable models even in settings with weak signals, limited dataset size, and limited compute.

• Pavlović, Milena; Hajj, Ghadi al; Kanduri, Chakravarthi; Pensar, Johan; Wood, Mollie Elizabeth & Sollid, Ludvig Magne [Show all 8 contributors for this article] (2024). Improving generalization of machine learning-identified biomarkers using causal modelling with examples from immune receptor diagnostics. Nature Machine Intelligence. 6(1), p. 15–24. doi: 10.1038/s42256-023-00781-8. Full text in Research Archive
• Kuronen, Juri; Horsfield, Samuel T.; Pöntinen, Anna Kaarina; Mallawaarachchi, Sudaraka; Alonso, Sergio Arredondo & Thorpe, Harry [Show all 14 contributors for this article] (2024). Pangenome-spanning epistasis and coselection analysis via de Bruijn graphs. Genome Research. ISSN 1088-9051. 34(7), p. 1081–1088. doi: 10.1101/gr.278485.123. Full text in Research Archive Show summary

  Studies of bacterial adaptation and evolution are hampered by the difficulty of measuring traits such as virulence, drug resistance, and transmissibility in large populations. In contrast, it is now feasible to obtain high-quality complete assemblies of many bacterial genomes thanks to scalable high-accuracy long-read sequencing technologies. To exploit this opportunity, we introduce a phenotype- and alignment-free method for discovering coselected and epistatically interacting genomic variation from genome assemblies covering both core and accessory parts of genomes. Our approach uses a compact colored de Bruijn graph to approximate the intragenome distances between pairs of loci for a collection of bacterial genomes to account for the impacts of linkage disequilibrium (LD). We demonstrate the versatility of our approach to efficiently identify associations between loci linked with drug resistance and adaptation to the hospital niche in the major human bacterial pathogens Streptococcus pneumoniae and Enterococcus faecalis.

• Pascoe, Ben; Futcher, Georgina; Pensar, Johan; Bayliss, Sion C.; Mourkas, Evangelos & Calland, Jessica Kate [Show all 21 contributors for this article] (2024). Machine learning to attribute the source of Campylobacter infections in the United States: A retrospective analysis of national surveillance data. Journal of Infection. ISSN 0163-4453. 89(5). doi: 10.1016/j.jinf.2024.106265. Full text in Research Archive Show summary

  Objectives Integrating pathogen genomic surveillance with bioinformatics can enhance public health responses by identifying risk and guiding interventions. This study focusses on the two predominant Campylobacter species, which are commonly found in the gut of birds and mammals and often infect humans via contaminated food. Rising incidence and antimicrobial resistance (AMR) are a global concern, and there is an urgent need to quantify the main routes to human infection. Methods During routine US national surveillance (2009–2019), 8856 Campylobacter genomes from human infections and 16,703 from possible sources were sequenced. Using machine learning and probabilistic models, we target genetic variation associated with host adaptation to attribute the source of human infections and estimate the importance of different disease reservoirs. Results Poultry was identified as the primary source of human infections, responsible for an estimated 68% of cases, followed by cattle (28%), and only a small contribution from wild birds (3%) and pork sources (1%). There was also evidence of an increase in multidrug resistance, particularly among isolates attributed to chickens. Conclusions National surveillance and source attribution can guide policy, and our study suggests that interventions targeting poultry will yield the greatest reductions in campylobacteriosis and spread of AMR in the US. Data availability All sequence reads were uploaded and shared on NCBI’s Sequence Read Archive (SRA) associated with BioProjects; PRJNA239251 (CDC / PulseNet surveillance), PRJNA287430 (FSIS surveillance), PRJNA292668 & PRJNA292664 (NARMS) and PRJNA258022 (FDA surveillance). Publicly available genomes, including reference genomes and isolates sampled worldwide from wild birds are associated with BioProject accessions: PRJNA176480, PRJNA177352, PRJNA342755, PRJNA345429, PRJNA312235, PRJNA415188, PRJNA524300, PRJNA528879, PRJNA529798, PRJNA575343, PRJNA524315 and PRJNA689604. Contiguous assemblies of all genome sequences compared are available at Mendeley data (assembled C. coli genomes doi: 10.17632/gxswjvxyh3.1; assembled C. jejuni genomes doi: 10.17632/6ngsz3dtbd.1) and individual project and accession numbers can be found in Supplementary tables S1 and S2, which also includes pubMLST identifiers for assembled genomes. Figshare (10.6084/m9.figshare.20279928). Interactive phylogenies are hosted on microreact separately for C. jejuni (https://microreact.org/project/pascoe-us-cjejuni) and C. coli (https://microreact.org/project/pascoe-us-ccoli).

• Hjort, Anders Dahl; Williams, Jonathan P. & Pensar, Johan (2024). Clustered Conformal Prediction for the Housing Market. Proceedings of Machine Learning Research (PMLR). 230, p. 366–386. Full text in Research Archive
• Hjort, Anders Dahl; Scheel, Ida; Sommervoll, Dag Einar & Pensar, Johan (2023). Locally interpretable tree boosting: An application to house price prediction. Decision Support Systems. ISSN 0167-9236. 178. doi: 10.1016/j.dss.2023.114106. Full text in Research Archive Show summary

  We introduce Locally Interpretable Tree Boosting (LitBoost), a tree boosting model tailored to applications where the data comes from several heterogeneous yet known groups with a limited number of observations per group. LitBoost constraints the complexity of a Gradient Boosted Trees model in a way that allows us to express the final model as a set of local Generalized Additive Models, yielding significant interpretability benefits while still maintaining some of the predictive power of a Gradient Boosted Trees model. We use house price prediction as a motivating example and demonstrate the performance of LitBoost on a data set of observations from different city districts in Oslo (Norway). We also test the robustness of LitBoost in an extensive simulation study on a synthetic data set.

• Hajj, Ghadi al; Pensar, Johan & Sandve, Geir Kjetil Ferkingstad (2023). DagSim: Combining DAG-based model structure with unconstrained data types and relations for flexible, transparent, and modularized data simulation. PLOS ONE. 18(4). doi: 10.1371/journal.pone.0284443. Full text in Research Archive Show summary

  Data simulation is fundamental for machine learning and causal inference, as it allows exploration of scenarios and assessment of methods in settings with full control of ground truth. Directed acyclic graphs (DAGs) are well established for encoding the dependence structure over a collection of variables in both inference and simulation settings. However, while modern machine learning is applied to data of an increasingly complex nature, DAG-based simulation frameworks are still confined to settings with relatively simple variable types and functional forms. We here present DagSim, a Python-based framework for DAG-based data simulation without any constraints on variable types or functional relations. A succinct YAML format for defining the simulation model structure promotes transparency, while separate user-provided functions for generating each variable based on its parents ensure simulation code modularization. We illustrate the capabilities of DagSim through use cases where metadata variables control shapes in an image and patterns in bio-sequences. DagSim is available as a Python package at PyPI. Source code and documentation are available at: https://github.com/uio-bmi/dagsim

• Corander, Jukka; Hanage, William P & Pensar, Johan (2022). Causal discovery for the microbiome. Lancet Microbe. 3(11), p. e881–e887. doi: 10.1016/S2666-5247(22)00186-0. Full text in Research Archive Show summary

  Measurement and manipulation of the microbiome is generally considered to have great potential for understanding the causes of complex diseases in humans, developing new therapies, and finding preventive measures. Many studies have found significant associations between the microbiome and various diseases; however, Koch's classical postulates remind us about the importance of causative reasoning when considering the relationship between microbes and a disease manifestation. Although causal discovery in observational microbiome data faces many challenges, methodological advances in causal structure learning have improved the potential of data-driven prediction of causal effects in large-scale biological systems. In this Personal View, we show the capability of existing methods for inferring causal effects from metagenomic data, and we highlight ways in which the introduction of causal structures that are more flexible than existing structures offers new opportunities for causal reasoning. Our observations suggest that microbiome research can further benefit from tools developed in the past 5 years in causal discovery and learn from their applications elsewhere.

• Hjort, Anders Dahl; Pensar, Johan; Scheel, Ida & Sommervoll, Dag Einar (2022). House price prediction with gradient boosted trees under different loss functions. Journal of Property Research. ISSN 0959-9916. 39(4), p. 333–364. doi: 10.1080/09599916.2022.2070525. Full text in Research Archive Show summary

  Many banks and credit institutions are required to assess the value of dwellings in their mortgage portfolio. This valuation often relies on an Automated Valuation Model (AVM). Moreover, these institutions often report the models accuracy by two numbers: The fraction of predictions within ±20% and ±10% range from the true values. Until recently, AVMs tended to be hedonic regression models, but lately machine learning approaches like random forest and gradient boosted trees have been increasingly applied. Both the traditional approaches and the machine learning approaches rely on minimising mean squared prediction error, and not the number of predictions in the ±20% and ±10% range. We investigate whether introducing a loss function closer to the AVMs actual loss measure improves performance in machine learning approaches, specifically for a gradient boosted tree approach. This loss function yields an improvement from 89.4% to 90.0% of predictions within ±20% of the true value on a data set of N=126719 transactions from the Norwegian housing market between 2013 and 2015, with the biggest improvements in performance coming from the lower price segments. We also find that a weighted average of models with different loss functions improves performance further, yielding 90.4% of the observations within ±20% of the true value.

• Chewapreecha, Claire; Pensar, Johan; Chattagul, Supaksorn; Pesonen, Maiju; Sangphukieo, Apiwat & Boonklang, Phumrapee [Show all 18 contributors for this article] (2021). Co-evolutionary Signals Identify Burkholderia pseudomallei Survival Strategies in a Hostile Environment. Molecular Biology and Evolution (MBE). ISSN 0737-4038. 39(1). doi: 10.1093/molbev/msab306. Full text in Research Archive Show summary

  Abstract The soil bacterium Burkholderia pseudomallei is the causative agent of melioidosis and a significant cause of human morbidity and mortality in many tropical and subtropical countries. The species notoriously survives harsh environmental conditions but the genetic architecture for these adaptations remains unclear. Here we employed a powerful combination of genome-wide epistasis and co-selection studies (2,011 genomes), condition-wide transcriptome analyses (82 diverse conditions), and a gene knockout assay to uncover signals of “co-selection”—that is a combination of genetic markers that have been repeatedly selected together through B. pseudomallei evolution. These enabled us to identify 13,061 mutation pairs under co-selection in distinct genes and noncoding RNA. Genes under co-selection displayed marked expression correlation when B. pseudomallei was subjected to physical stress conditions, highlighting the conditions as one of the major evolutionary driving forces for this bacterium. We identified a putative adhesin (BPSL1661) as a hub of co-selection signals, experimentally confirmed a BPSL1661 role under nutrient deprivation, and explored the functional basis of co-selection gene network surrounding BPSL1661 in facilitating the bacterial survival under nutrient depletion. Our findings suggest that nutrient-limited conditions have been the common selection pressure acting on this species, and allelic variation of BPSL1661 may have promoted B. pseudomallei survival during harsh environmental conditions by facilitating bacterial adherence to different surfaces, cells, or living hosts.

• Suotsalo, Kimmo; Xu, Yingying; Corander, Jukka & Pensar, Johan (2021). High-dimensional structure learning of sparse vector autoregressive models using fractional marginal pseudo-likelihood. Statistics and computing. ISSN 0960-3174. 31(73). doi: 10.1007/s11222-021-10049-z. Full text in Research Archive Show summary

  Abstract Learning vector autoregressive models from multivariate time series is conventionally approached through least squares or maximum likelihood estimation. These methods typically assume a fully connected model which provides no direct insight to the model structure and may lead to highly noisy estimates of the parameters. Because of these limitations, there has been an increasing interest towards methods that produce sparse estimates through penalized regression. However, such methods are computationally intensive and may become prohibitively time-consuming when the number of variables in the model increases. In this paper we adopt an approximate Bayesian approach to the learning problem by combining fractional marginal likelihood and pseudo-likelihood. We propose a novel method, PLVAR, that is both faster and produces more accurate estimates than the state-of-the-art methods based on penalized regression. We prove the consistency of the PLVAR estimator and demonstrate the attractive performance of the method on both simulated and real-world data.

• Mageiros, Leonardos; Meric, Guillaume; Bayliss, Sion; Pensar, Johan; Pascoe, Ben & Mourkas, Evangelos [Show all 19 contributors for this article] (2021). Genome evolution and the emergence of pathogenicity in avian Escherichia coli. Nature Communications. 12(1). doi: 10.1038/s41467-021-20988-w. Full text in Research Archive Show summary

  Abstract Chickens are the most common birds on Earth and colibacillosis is among the most common diseases affecting them. This major threat to animal welfare and safe sustainable food production is difficult to combat because the etiological agent, avian pathogenic Escherichia coli (APEC), emerges from ubiquitous commensal gut bacteria, with no single virulence gene present in all disease-causing isolates. Here, we address the underlying evolutionary mechanisms of extraintestinal spread and systemic infection in poultry. Combining population scale comparative genomics and pangenome-wide association studies, we compare E. coli from commensal carriage and systemic infections. We identify phylogroup-specific and species-wide genetic elements that are enriched in APEC, including pathogenicity-associated variation in 143 genes that have diverse functions, including genes involved in metabolism, lipopolysaccharide synthesis, heat shock response, antimicrobial resistance and toxicity. We find that horizontal gene transfer spreads pathogenicity elements, allowing divergent clones to cause infection. Finally, a Random Forest model prediction of disease status (carriage vs. disease) identifies pathogenic strains in the emergent ST-117 poultry-associated lineage with 73% accuracy, demonstrating the potential for early identification of emergent APEC in healthy flocks.

• Pavlović, Milena; Scheffer, Lonneke; Motwani, Keshav; Kanduri, Chakravarthi; Kompova, Radmila & Vazov, Nikolay Aleksandrov [Show all 42 contributors for this article] (2021). The immuneML ecosystem for machine learning analysis of adaptive immune receptor repertoires. Nature Machine Intelligence. 3(11), p. 936–944. doi: 10.1038/s42256-021-00413-z. Full text in Research Archive
• Viinikka, Jussi; Hyttinen, Antti; Pensar, Johan & Koivisto, Mikko (2020). Towards Scalable Bayesian Learning of Causal DAGs. Advances in Neural Information Processing Systems. ISSN 1049-5258. Full text in Research Archive
• Top, Janetta; Arredondo-Alonso, Sergio; Schürch, Anita C.; Puranen, Santeri; Pesonen, Maiju & Pensar, Johan [Show all 8 contributors for this article] (2020). Genomic rearrangements uncovered by genome-wide co-evolution analysis of a major nosocomial pathogen, Enterococcus faecium. Microbial Genomics. 6(12), p. 1–8. doi: 10.1099/mgen.0.000488. Full text in Research Archive Show summary

  Enterococcus faecium is a gut commensal of the gastro-digestive tract, but also known as nosocomial pathogen among hospitalized patients. Population genetics based on whole-genome sequencing has revealed that E. faecium strains from hospitalized patients form a distinct clade, designated clade A1, and that plasmids are major contributors to the emergence of nosocomial E. faecium . Here we further explored the adaptive evolution of E. faecium using a genome-wide co-evolution study (GWES) to identify co-evolving single-nucleotide polymorphisms (SNPs). We identified three genomic regions harbouring large numbers of SNPs in tight linkage that are not proximal to each other based on the completely assembled chromosome of the clade A1 reference hospital isolate AUS0004. Close examination of these regions revealed that they are located at the borders of four different types of large-scale genomic rearrangements, insertion sites of two different genomic islands and an IS30-like transposon. In non-clade A1 isolates, these regions are adjacent to each other and they lack the insertions of the genomic islands and IS30-like transposon. Additionally, among the clade A1 isolates there is one group of pet isolates lacking the genomic rearrangement and insertion of the genomic islands, suggesting a distinct evolutionary trajectory. In silico analysis of the biological functions of the genes encoded in three regions revealed a common link to a stress response. This suggests that these rearrangements may reflect adaptation to the stringent conditions in the hospital environment, such as antibiotics and detergents, to which bacteria are exposed. In conclusion, to our knowledge, this is the first study using GWES to identify genomic rearrangements, suggesting that there is considerable untapped potential to unravel hidden evolutionary signals from population genomic data.

• Tadei, Alessandro; Haajanen, Juulia; Pensar, Johan; Santtila, Pekka & Antfolk, Jan (2020). Counteracting deceptive responding in the Finnish Investigative Instrument of Child Sexual Abuse (FICSA). Journal of Sexual Aggression. ISSN 1355-2600. doi: 10.1080/13552600.2020.1846802. Full text in Research Archive Show summary

  Unfounded child sexual abuse (CSA) allegations take investigative resources from real cases and have detrimental consequences for the people involved. The Finnish Investigative Instrument of Child Sexual Abuse (FICSA) supports investigators by estimating the probability of a CSA allegation being true based on the child’s background information. In the current study, we aimed at making FICSA resistant to deception. Two gender-specific questionnaires with FICSA questions and additional “trap” questions were constructed. The trap questions were designed to seem statistically related to CSA although they were not. Combining the answers of 278 real victims and 275 16–year-old students, instructed to simulate being CSA victims, we built a Naïve Bayes classifier that successfully separated the two groups (AUC = 0.91 for boys and 0.92 for girls). By identifying false allegations early in the investigation, authorities’ resources can be directed towards allegations that are more likely true, effectively helping actual CSA victims.

View all works in NVA

• Kvisgaard, Vera Haugen & Pensar, Johan (2025). Bayesian IDA with Background Knowledge. doi: https:/nva.sikt.no/registration/019928073510-ddb0c9ca-acbc-44f5-8c56-f5a5720d1092. Full text in Research Archive Show summary

  There have been several recent advancements in the development of causal learning methods that combine structure learning and causal inference. Inspired by the frequentist IDA method, the Bayesian IDA (BIDA) method is a scalable Bayesian approach for estimating causal effects from observational data under an unknown causal Bayesian network. Here, we extend BIDA to the setting where the causal structure is partly known.

• Pensar, Johan (2024). Bayesian estimation of intervention effects using directed graphical models. Full text in Research Archive
• Aastveit, Marthe Elisabeth; Thorarinsdottir, Thordis Linda; Lenkoski, Alex & Pensar, Johan (2024). Demand changes over time in the short-term rental markets. Full text in Research Archive
• Pérez, Johan Manuel de Aguas; Pensar, Johan; Pérez, Tomás Fernando Varnet & Biele, Guido (2024). Recovering target causal effects from post-exposure selection induced by missing outcome data. Full text in Research Archive
• Pérez, Johan Manuel de Aguas; Pensar, Johan & Biele, Guido (2024). Partial identification and efficient estimation for the stratum-specific probability of benefit with thresholds on a continuous outcome. Full text in Research Archive
• Pérez, Johan Manuel de Aguas; Pensar, Johan; Pérez, Tomás Fernando Varnet & Biele, Guido (2024). Recovering target causal effects from post-exposure selection induced by missing outcome data. Full text in Research Archive
• Pérez, Johan Manuel de Aguas; Pensar, Johan & Biele, Guido (2023). Recovering causal effects from confounding and selection biases: Extended graphical criteria via mediation analysis. Full text in Research Archive
• Pérez, Johan Manuel de Aguas; Pensar, Johan & Biele, Guido (2023). Recovering causal effects from confounding and selection biases: Extended graphical criteria via mediation analysis. Full text in Research Archive
• Pérez, Johan Manuel de Aguas; Pensar, Johan & Biele, Guido (2023). Counterfactual inference with nonparametric Bayes: Causal attribution of ADHD treatment on level achievement at the national test. Full text in Research Archive
• Meggio, Claudio; Balcells, David; Pensar, Johan & Bin, Riccardo De (2023). Path Boosting, a statistical learning method for chemistry applications. Full text in Research Archive
• Meggio, Claudio; Balcells, David; Bin, Riccardo De & Pensar, Johan (2023). Adapting Gradient Booting methods to graph inputs. Full text in Research Archive
• Sandve, Geir Kjetil Ferkingstad & Pensar, Johan (2022). Machine Learning and Causality. Full text in Research Archive
• Mageiros, Leonardos; Meric, Guillaume; Bayliss, Sion; Pensar, Johan; Pascoe, Ben & Mourkas, Evangelos [Show all 19 contributors for this article] (2021). Author Correction: Genome evolution and the emergence of pathogenicity in avian Escherichia coli (Nature Communications, (2021), 12, 1, (765), 10.1038/s41467-021-20988-w). Nature Communications. 12(1). doi: 10.1038/s41467-021-22238-5. Full text in Research Archive

View all works in NVA