Wesley Schaal

The ability to analyze whether DNA contains lesions is essential in identifying mutagenic substances. Currently, the detection of single-stranded DNA breaks (SSBs) lacks precision. To address this limitation, we develop a method for sequence-templated erroneous end-labelling sequencing (STEEL-seq), which enables the mapping of SSBs. The method requires a highly error-prone DNA polymerase, so we engineer a chimeric DNA polymerase, Sloppymerase, capable of replicating DNA in the absence of one nucleotide. Following the omission of a specific nucleotide (e.g., dATP) from the reaction mixture, Sloppymerase introduces mismatches directly downstream of SSBs at positions where deoxyadenosine should occur. This mismatch pattern, coupled with the retention of sequence information flanking these sites, ensures that the identified hits are bona fide SSBs. STEEL-seq is compatible with a variety of sequencing technologies, as demonstrated using Sanger, Illumina, PacBio, and Nanopore systems. Using STEEL-seq, we determine the SSB/base pair frequency in the human genome to range between 0.7 and 3.8 × 10⁻⁶ with an enrichment in active promoter regions.

Comparing chemical structures to infer protein targets and functions is a common approach, but basing comparisons on chemical similarity alone can be misleading. Here we present a methodology for predicting target protein clusters using deep neural networks. The model is trained on clusters of compounds based on similarities calculated from combined compound-protein and protein-protein interaction data using a network topology approach. We compare several deep learning architectures including both convolutional and recurrent neural networks. The best performing method, the recurrent neural network architecture MolPMoFiT, achieved an F1 score approaching 0.9 on a held-out test set of 8907 compounds. In addition, in-depth analysis on a set of eleven well-studied chemical compounds with known functions showed that predictions were justifiable for all but one of the chemicals. Four of the compounds, similar in their molecular structure but with dissimilarities in their function, revealed advantages of our method compared to using chemical similarity.

Background: Identifying and assessing ligand-target binding is a core component in early drug discovery as one or more unwanted interactions may be associated with safety issues.Contributions: We present an open-source, extendable web service for predicting target profiles with confidence using machine learning for a panel of 7 targets, where models are trained on molecular docking scores from a large virtual library. The method uses conformal prediction to produce valid measures of prediction efficiency for a particular confidence level. The service also offers the possibility to dock chemical structures to the panel of targets with QuickVina on individual compound basis.Results: The docking procedure and resulting models were validated by docking well-known inhibitors for each of the 7 targets using QuickVina. The model predictions showed comparable performance to molecular docking scores against an external validation set. The implementation as publicly available microservices on Kubernetes ensures resilience, scalability, and extensibility.

Background: Identifying and assessing ligand-target binding is a core component in early drug discovery asone or more unwanted interactions may be associated with safety issues.Contributions: We present an open-source, extendable web service for predicting target profiles withconfidence using machine learning for a panel of 7 targets, where models are trained on molecular dockingscores from a large virtual library. The method uses conformal prediction to produce valid measures ofprediction efficiency for a particular confidence level. The service also offers the possibility to dock chemicalstructures to the panel of targets with QuickVina on individual compound basis.Results: The docking procedure and resulting models were validated by docking well-known inhibitors for eachof the 7 targets using QuickVina. The model predictions showed comparable performance to molecular dockingscores against an external validation set. The implementation as publicly available microservices on Kubernetesensures resilience, scalability, and extensibility

By collecting and analyzing diapers, we identified a >6-fold increase in carriage of extended-spectrum beta-lactamase (ESBL)-producing Enterobacteriaceae for healthy preschool children in Sweden (p= 40%. We analyzed samples from 334 children and found 56 containing >= 1 ESBL producer. The prevalence in the study population increased from 2.6% in 2010 to 16.8% in 2016 (p= 40%. Furthermore, 58% of the ESBL producers were multidrug resistant, and transmission of ESBL-producing and non-ESBL-producing strains was observed at several of the preschools. Toddlers appear to be major carriers of ESBL producers in Sweden.

Background: Next-generation sequencing (NGS) has transformed the life sciences, and many research groups are newly dependent upon computer clusters to store and analyze large datasets. This creates challenges for e-infrastructures accustomed to hosting computationally mature research in other sciences. Using data gathered from our own clusters at UPPMAX computing center at Uppsala University, Sweden, where core hour usage of ~800 NGS and ~200 non-NGS projects is now similar, we compare and contrast the growth, administrative burden, and cluster usage of NGS projects with projects from other sciences. Results: The number of NGS projects has grown rapidly since 2010, with growth driven by entry of new research groups. Storage used by NGS projects has grown more rapidly since 2013 and is now limited by disk capacity. NGS users submit nearly twice as many support tickets per user, and 11 more tools are installed each month for NGS projects than for non-NGS projects. We developed usage and efficiency metrics and show that computing jobs for NGS projects use more RAM than non-NGS projects, are more variable in core usage, and rarely span multiple nodes. NGS jobs use booked resources less efficiently for a variety of reasons. Active monitoring can improve this somewhat. Conclusions: Hosting NGS projects imposes a large administrative burden at UPPMAX due to large numbers of inexperienced users and diverse and rapidly evolving research areas. We provide a set of recommendations for e-infrastructures that host NGS research projects. We provide anonymized versions of our storage, job, and efficiency databases.

Docking and scoring large libraries of ligands against target proteins forms the basis of structure-based virtual screening. The problem is trivially parallelizable, and calculations are generally carried out on computer clusters or on large workstations in a brute force manner, by docking and scoring all available ligands. In this study we propose a strategy that is based on iteratively docking a set of ligands to form a training set, training a ligand-based model on this set, and predicting the remainder of the ligands to exclude those predicted as ‘low-scoring’ ligands. Then, another set of ligands are docked, the model is retrained and the process is repeated until a certain model efficiency level is reached. Thereafter, the remaining ligands are docked or excluded based on this model. We use SVM and conformal prediction to deliver valid prediction intervals for ranking the predicted ligands, and Apache Spark to parallelize both the docking and the modeling. We show on 4 different targets that conformal prediction based virtual screening (CPVS) is able to reduce the number of docked molecules by 62.61% while retaining an accuracy for the top 30 hits of 94% on average and a speedup of 3.7. The implementation is available as open source via GitHub ( https://github.com/laeeq80/spark-cpvs ) and can be run on high-performance computers as well as on cloud resources.