Volume 20, Issue 2 - Special Section: Selected extended papers of the 16th International Conference on Practical Applications of Computational Biology and Bioinformatics (PACBB); Guest Editor: Florentino Fdez-Riverola

EvoPPI (http://evoppi.i3s.up.pt), a meta-database for protein-protein interactions (PPI), has been upgraded (EvoPPI3) to accept new types of data, namely, PPI from patients, cell lines, and animal models, as well as data from gene modifier experiments, for nine neurodegenerative polyglutamine (polyQ) diseases caused by an abnormal expansion of the polyQ tract. The integration of the different types of data allows users to easily compare them, as here shown for Ataxin-1, the polyQ protein involved in spinocerebellar ataxia type 1 (SCA1) disease. Using all available datasets and the data here obtained for Drosophila melanogaster wt and exp Ataxin-1 mutants (also available at EvoPPI3), we show that, in humans, the Ataxin-1 network is much larger than previously thought (380 interactors), with at least 909 interactors. The functional profiling of the newly identified interactors is similar to the ones already reported in the main PPI databases. 16 out of 909 interactors are putative novel SCA1 therapeutic targets, and all but one are already being studied in the context of this disease. The 16 proteins are mainly involved in binding and catalytic activity (mainly kinase activity), functional features already thought to be important in the SCA1 disease.

To support physicians in clinical decision process on patients affected by Coronavirus Disease 2019 (COVID-19) in areas with a low vaccination rate, we devised and evaluated the performances of several machine learning (ML) classifiers fed with readily available clinical and laboratory data. Our observational retrospective study collected data from a cohort of 779 COVID-19 patients presenting to three hospitals of the Lazio-Abruzzo area (Italy). Based on a different selection of clinical and respiratory (ROX index and PaO2/FiO2 ratio) variables, we devised an AI-driven tool to predict safe discharge from ED, disease severity and mortality during hospitalization. To predict safe discharge our best classifier is an RF integrated with ROX index that reached AUC of 0.96. To predict disease severity the best classifier was an RF integrated with ROX index that reached an AUC of 0.91. For mortality prediction the best classifier was an RF integrated with ROX index, that reached an AUC of 0.91. The results obtained thanks to our algorithms are consistent with the scientific literature an accomplish significant performances to forecast safe discharge from ED and severe clinical course of COVID-19.

Nicotinamide adenine dinucleotide (NAD) levels are essential for the normal physiology of the cell and are strictly regulated to prevent pathological conditions. NAD functions as a coenzyme in redox reactions, as a substrate of regulatory proteins, and as a mediator of protein-protein interactions. The main objectives of this study were to identify the NAD-binding and NAD-interacting proteins, and to uncover novel proteins and functions that could be regulated by this metabolite. It was considered if cancer-associated proteins were potential therapeutic targets. Using multiple experimental databases, we defined datasets of proteins that directly interact with NAD – the NAD-binding proteins ( NADBPs ) dataset – and of proteins that interact with NADBPs – the NAD-protein–protein interactions ( NAD-PPIs ) dataset. Pathway enrichment analysis revealed that NADBPs participate in several metabolic pathways, while NAD-PPIs are mostly involved in signalling pathways. These include disease-related pathways, namely, three major neurodegenerative disorders: Alzheimer’s disease, Huntington’s disease, and Parkinson’s disease. Then, the complete human proteome was further analysed to select potential NADBPs. TRPC3 and isoforms of diacylglycerol (DAG) kinases, which are involved in calcium signalling, were identified as new NADBPs. Potential therapeutic targets that interact with NAD were identified, that have regulatory and signalling functions in cancer and neurodegenerative diseases.

NDM-1 (New-Delhi-Metallo-β-lactamase-1) is an enzyme developed by bacteria that is implicated in bacteria resistance to almost all known antibiotics. In this study, we deliver a new, curated NDM-1 bioactivities database, along with a set of unifying rules for managing different activity properties and inconsistencies. We define the activity classification problem in terms of Multiple Instance Learning, employing embeddings corresponding to molecular substructures and present an ensemble ranking and classification framework, relaying on a k-fold Cross Validation method employing a per fold hyper-parameter optimization procedure, showing promising generalization ability. The MIL paradigm displayed an improvement up to 45.7 %, in terms of Balanced Accuracy, in comparison to the classical Machine Learning paradigm. Moreover, we investigate different compact molecular representations, based on atomic or bi-atomic substructures. Finally, we scanned the Drugbank for strongly active compounds and we present the top-15 ranked compounds.

Analyzing metabolic pathways in systems biology requires accurate kinetic parameters that represent the simulated in vivo processes. Simulation of the fermentation pathway in the Saccharomyces cerevisiae kinetic model help saves much time in the optimization process. Fitting the simulated model into the experimental data is categorized under the parameter estimation problem. Parameter estimation is conducted to obtain the optimal values for parameters related to the fermentation process. This step is essential because insufficient identification of model parameters can cause erroneous conclusions. The kinetic parameters cannot be measured directly. Therefore, they must be estimated from the experimental data either in vitro or in vivo . Parameter estimation is a challenging task in the biological process due to the complexity and nonlinearity of the model. Therefore, we propose the Artificial Bee Colony algorithm (ABC) to estimate the parameters in the fermentation pathway of S. cerevisiae to obtain more accurate values. A metabolite with a total of six parameters is involved in this article. The experimental results show that ABC outperforms other estimation algorithms and gives more accurate kinetic parameter values for the simulated model. Most of the estimated kinetic parameter values obtained from the proposed algorithm are the closest to the experimental data.

This work aims to describe the observed enrichment of inverted repeats in the human genome; and to identify and describe, with detailed length profiles, the regions with significant and relevant enriched occurrence of inverted repeats. The enrichment is assessed and tested with a recently proposed measure ( z -scores based measure). We simulate a genome using an order 7 Markov model trained with the data from the real genome. The simulated genome is used to establish the critical values which are used as decision thresholds to identify the regions with significant enriched concentrations. Several human genome regions are highly enriched in the occurrence of inverted repeats. This is observed in all the human chromosomes. The distribution of inverted repeat lengths varies along the genome. The majority of the regions with severely exaggerated enrichment contain mainly short length inverted repeats. There are also regions with regular peaks along the inverted repeats lengths distribution (periodic regularities) and other regions with exaggerated enrichment for long lengths (less frequent). However, adjacent regions tend to have similar distributions.

Transmembrane transport proteins (transporters) play a crucial role in the fundamental cellular processes of all organisms by facilitating the transport of hydrophilic substrates across hydrophobic membranes. Despite the availability of numerous membrane protein sequences, their structures and functions remain largely elusive. Recently, natural language processing (NLP) techniques have shown promise in the analysis of protein sequences. Bidirectional Encoder Representations from Transformers (BERT) is an NLP technique adapted for proteins to learn contextual embeddings of individual amino acids within a protein sequence. Our previous strategy, TooT-BERT-T, differentiated transporters from non-transporters by employing a logistic regression classifier with fine-tuned representations from ProtBERT-BFD. In this study, we expand upon this approach by utilizing representations from ProtBERT, ProtBERT-BFD, and MembraneBERT in combination with classical classifiers. Additionally, we introduce TooT-BERT-CNN-T, a novel method that fine-tunes ProtBERT-BFD and discriminates transporters using a Convolutional Neural Network (CNN). Our experimental results reveal that CNN surpasses traditional classifiers in discriminating transporters from non-transporters, achieving an MCC of 0.89 and an accuracy of 95.1 % on the independent test set. This represents an improvement of 0.03 and 1.11 percentage points compared to TooT-BERT-T, respectively.