Posters - Schedules

Introduction: Most antiarrhythmic drugs to treat atrial fibrillation (AF) exhibit multi-target action on ion channels. However, the structural basis of their promiscuity is not well-understood. Here we compared the binding sites (BS) of two antiarrhythmic drugs, flecainide and AVE0118, in potassium Kv1.5 and sodium Nav1.5 ion channels. The latter are involved in cardiac action potential and constitute relevant targets for AF. Methods: We performed molecular docking to predict flecainide and AVE0118 BS in Kv1.5 and Nav1.5, respectively. BS comparisons were done using an in-house workflow combining molecular dynamics, BS characterization, and pattern matching. Ala-mutations in Nav1.5 wild-type were obtained by restriction enzyme cloning and their biophysical characterization by whole-cell patch clamp. Results and Discussion: We predicted that flecainide and AVE0118 drugs occupy the central cavity. Besides, flecainide extends to the inter-subunit interfase in Kv1.5 while AVE0118 inserts into fenestrations in Nav1.5. The comparison of BS for AVE0118 is pending to complete Ala-mutations in Nav1.5 to validate our docking predictions; at present we confirmed F1465A mutation by DNA sequencing. Regarding flecainide, previous structural and mutagenesis data coupled with our BS comparison workflow were sufficient to identify a common structural pattern in Nav1.5 and Kv1.5. The latter belongs to the central cavity and consists of a hydrophobic patch and a polar region. We also identified a distinctive feature of aromatic residues in flecainide BS in Nav1.5, probably linked to its higher affinity. Our findings could advance the understanding of drug promiscuity in ion channels, and give insights for multi-target directed ligand design with applications to AF.

More and more evolutionary and molecular biologists are interested in building alternative supertrees. Often, developing new approaches or testing new metrics requires relevant datasets that is not always easy to obtain. In order to solve this problem of lack of data, we propose a new approach and developed a program in Python to generate overlapping phylogenetic trees with biological events to simplify the process of obtaining this type of data.
The new tool takes the number of phylogenetic trees the user wants to generate, the maximum number of leaves per tree to generate, and the average level of leaf overlap between phylogenetic trees as input parameters. The program returns to the user a set of phylogenetic trees in Newick format, respecting the parameters given as input, in order to use them to infer a supertree (or supertrees). This data can be an important resource for research; the user can download the generated data and use it later in their relevant application tasks.

Third-generation long read sequencing platforms still introduce relatively high error rates, despite their rapid improvement in accuracy since their inception. PCR amplification also introduces sequence error. It is critical to correctly accommodate the error profiles introduced to sequences of interest when clustering them. Examples of such sequences include UMIs, barcodes, motifs, and transcription factor binding sites.
Levenshtein distance (LD) calculates the number of edits required to make one string match another. Clustering related biological sequences is dependent on establishing a threshold of similarity based on edit distance. Previous attempts at modifying Levenshtein distance for biological clustering have provided some optimization but have fallen short. However, we have recently been successful in modifying LD to cluster DNA sequences while accommodating weights assigned to insertions, deletions, and substitutions based on their frequency of appearance and allowing for frameshift accommodations. This present work presents the underlying algorithm behind our recent improvement.

Antibiotic resistance is considered one of the most urgent threats to global public health.Due to the public health risk, there are several methods for obtained phenotypic results.However, conventional methods take days or weeks. Whole-genome sequencing (WGS)overcomes these limitations by estimating phenotypic behavior and identifying antibiotic resistance elements in the genome in a faster way. However, information about the optimal prediction of these profiles is still scarce. The project aim was to implement an antibiotic resistance prediction model using Machine Learning methods, using WGS data of 521 Enterobacterales isolates, including 28 Providencia rettgeri isolates sequenced in Colombia. The Machine Learning methods used were a) Logistic Regression (RL), b) Support Vector Machine (SVM), and c) Random Forest (RF). Also, the following feature selection methods were applied: 1) recursive feature elimination (RFECV), 2) L1 regularization, and 3) feature importance. Finally, prediction models were developed for 10 antibiotics, with a mean accuracy of 88% (IC 95% ± 6) and individual accuracies of 89% (IC 95% ± 7), 93% (IC 95% ± 5), 90% (IC 95% ± 7), 93% (IC 95% ± 6), 81% (IC 95% ± 12),93% (IC 95% ± 8) 81% (IC 95% ± 10), 79% (IC 95% ± 9), 86% (IC 95% ± 9) and 93% (IC 95% ± 5), for amikacin, ciprofloxacin, trimethoprim/sulfamethoxazole, tetracycline,tigecycline, colistin, ceftazidime, cefepime, imipenem and meropenem respectively. These performances correspond to RL and SVM, using RFECV and L1 as regularization feature selection methods. These findings indicate that these models could accurately predict antibiotic resistance from different Enterobacteriaceae species and could be a potential tool for clinical diagnosis.