Volume 14, Issue 4 - Special Issue: Selected extended papers of the 11th International Conference on Practical Applications of Computational Biology and Bioinformatics (PACBB) / Guest Editors: Florentino Fdez-Riverola, Miguel Rocha

Word sense disambiguation (WSD) is an important step in biomedical text mining, which is responsible for assigning an unequivocal concept to an ambiguous term, improving the accuracy of biomedical information extraction systems. In this work we followed supervised and knowledge-based disambiguation approaches, with the best results obtained by supervised means. In the supervised method we used bag-of-words as local features, and word embeddings as global features. In the knowledge-based method we combined word embeddings, concept textual definitions extracted from the UMLS database, and concept association values calculated from the MeSH co-occurrence counts from MEDLINE articles. Also, in the knowledge-based method, we tested different word embedding averaging functions to calculate the surrounding context vectors, with the goal to give more importance to closest words of the ambiguous term. The MSH WSD dataset, the most common dataset used for evaluating biomedical concept disambiguation, was used to evaluate our methods. We obtained a top accuracy of 95.6 % by supervised means, while the best knowledge-based accuracy was 87.4 %. Our results show that word embedding models improved the disambiguation accuracy, proving to be a powerful resource in the WSD task.

Curation of protein interactions from scientific articles is an important task, since interaction networks are essential for the understanding of biological processes associated with disease or pharmacological action for example. However, the increase in the number of publications that potentially contain relevant information turns this into a very challenging and expensive task. In this work we used a convolutional recurrent neural network for identifying relevant articles for extracting information regarding protein interactions. Using the BioCreative III Article Classification Task dataset, we achieved an area under the precision-recall curve of 0.715 and a Matthew’s correlation coefficient of 0.600, which represents an improvement over previous works.

Public resources need to be appropriately annotated with metadata in order to make them discoverable, reproducible and traceable, further enabling them to be interoperable or integrated with other datasets. While data-sharing policies exist to promote the annotation process by data owners, these guidelines are still largely ignored. In this manuscript, we analyse automatic measures of metadata quality, and suggest their application as a mean to encourage data owners to increase the metadata quality of their resources and submissions, thereby contributing to higher quality data, improved data sharing, and the overall accountability of scientific publications. We analyse these metadata quality measures in the context of a real-world repository of metabolomics data (i.e. MetaboLights), including a manual validation of the measures, and an analysis of their evolution over time. Our findings suggest that the proposed measures can be used to mimic a manual assessment of metadata quality.

Banana peels are well recognized as a source of important bioactive compounds, such as phenolics, carotenoids, biogenic amines, among others. As such, they have recently started to be used for industrial purposes. However, its composition seems to be strongly affected by biotic or abiotic ecological factors. Thus, this study aimed to investigate banana peels chemical composition, not only to get insights on eventual metabolic changes caused by the seasons, in southern Brazil, but also to identify the most relevant metabolites for these processes. To achieve this, a Nuclear magnetic resonance (NMR)-based metabolic profiling strategy was adopted, followed by chemometrics analysis, using the specmine package for the R environment, and metabolite identification. The results showed that the metabolomic approach adopted allowed identifying a series of primary and secondary metabolites in the aqueous extracts investigated. Besides, over the seasons the metabolic profiles of the banana peels showed to contain biologically active compounds relevant to the skin wound healing process, indicating the biotechnological potential of that raw material.

Vitamin A deficiency is a prevalent health problem in many areas of the world, where cassava genotypes with high pro-vitamin A content have been identified as a strategy to address this issue. In this study, we found a positive correlation between the color of the root pulp and the total carotenoid contents and, importantly, showed how CIELAB color measurements can be used as a non-destructive and fast technique to quantify the amount of carotenoids in cassava root samples, as opposed to traditional methods. We trained several machine learning models using UV-visible spectrophotometry data, CIELAB data and a low-level data fusion of the two. Best performance models were obtained for the total carotenoids contents calculated using the UV-visible dataset as input, with R 2 values above 90 %. Using CIELAB and fusion data, values around 60 % and above 90 % were found. Importantly, these results demonstrated how data fusion can lead to a better model performance for prediction when comparing to the use of a single data source. Considering all these findings, the use of colorimetric data associated with UV-visible and HPLC data through statistical and machine learning methods is a reliable way of predicting the content of total carotenoids in cassava root samples.

Novel transcript discovery through RNA sequencing has substantially improved our understanding of the transcriptome dynamics of biological systems. Endogenous target mimicry (eTM) transcripts, a novel class of regulatory molecules, bind to their target microRNAs (miRNAs) by base pairing and block their biological activity. The objective of this study was to provide a computational analysis framework for the prediction of putative eTM sequences in plants, and as an example, to discover previously un-annotated eTMs in Prunus persica (peach) transcriptome. Therefore, two public peach transcriptome libraries downloaded from Sequence Read Archive (SRA) and a previously published set of long non-coding RNAs (lncRNAs) were investigated with multi-step analysis pipeline, and 44 putative eTMs were found. Additionally, an eTM-miRNA-mRNA regulatory network module associated with peach fruit organ development was built via integration of the miRNA target information and predicted eTM-miRNA interactions. My findings suggest that one of the most widely expressed miRNA families among diverse plant species, miR156, might be potentially sponged by seven putative eTMs. Besides, the study indicates eTMs potentially play roles in the regulation of development processes in peach fruit via targeting specific miRNAs. In conclusion, by following the step-by step instructions provided in this study, novel eTMs can be identified and annotated effectively in public plant transcriptome libraries.

Planaria is a member of the Phylum Platyhelminthes including flatworms. Planarians possess the unique ability of regeneration from adult stem cells or neoblasts and finds importance as a model organism for regeneration and developmental studies. Although research is being actively carried out globally through conventional methods to understand the process of regeneration from neoblasts, biology of development, neurobiology and immunology of Planaria, there are many thought provoking questions related to stem cell plasticity, and uniqueness of regenerative potential in Planarians amongst other members of Phylum Platyhelminthes. The complexity of receptors and signalling mechanisms, immune system network, biology of repair, responses to injury are yet to be understood in Planaria. Genomic and transcriptomic studies have generated a vast repository of data, but their availability and analysis is a challenging task. Data mining, computational approaches of gene curation, bioinformatics tools for analysis of transcriptomic data, designing of databases, application of algorithms in deciphering changes of morphology by RNA interference (RNAi) approaches, understanding regeneration experiments is a new venture in Planaria research that is helping researchers across the globe in understanding the biology. We highlight the applications of Hidden Markov models (HMMs) in designing of computational tools and their applications in Planaria decoding their complex biology.