A variant of the glycoprotein CD44 (CD44v) that shares sequences with variants causally involved in metastasis formation is transiently expressed on B and T lymphocytes and macrophages after antigenic stimulation and in the postnatal period. Antibodies to the variant hinder in vivo activation of both B and T cells. The observation that a protein domain that is expressed on CD44 and required for the lymphatic spread of tumor cells can catalyze an essential step in the process of lymphocyte activation supports the idea that metastasizing tumor cells mimic lymphocyte behavior.

With this study, we provide a comprehensive reference dataset of detailed miRNA expression profiles from seven types of human peripheral blood cells (NK cells, B lymphocytes, cytotoxic T lymphocytes, T helper cells, monocytes, neutrophils and erythrocytes), serum, exosomes and whole blood. The peripheral blood cells from buffy coats were typed and sorted using FACS/MACS. The overall dataset was generated from 450 small RNA libraries using high-throughput sequencing. By employing a comprehensive bioinformatics and statistical analysis, we show that 3′ trimming modifications as well as composition of 3′ added non-templated nucleotides are distributed in a lineage-specific manner—the closer the hematopoietic progenitors are, the higher their similarities in sequence variation of the 3′ end. Furthermore, we define the blood cell-specific miRNA and isomiR expression patterns and identify novel cell type specific miRNA candidates. The study provides the most comprehensive contribution to date towards a complete miRNA catalogue of human peripheral blood, which can be used as a reference for future studies. The dataset has been deposited in GEO and also can be explored interactively following this link: http://134.245.63.235/ikmb-tools/bloodmiRs.

Deep neural networks (DNN) have fundamentally revolutionized the artificial intelligence (AI) field. The transformer model is a type of DNN that was originally used for the natural language processing tasks and has since gained more and more attention for processing various kinds of sequential data, including biological sequences and structured electronic health records. Along with this development, transformer-based models such as BioBERT, MedBERT, and MassGenie have been trained and deployed by researchers to answer various scientific questions originating in the biomedical domain. In this paper, we review the development and application of transformer models for analyzing various biomedical-related datasets such as biomedical textual data, protein sequences, medical structured-longitudinal data, and biomedical images as well as graphs. Also, we look at explainable AI strategies that help to comprehend the predictions of transformer-based models. Finally, we discuss the limitations and challenges of current models, and point out emerging novel research directions.

Motivation: Chemical compounds like small signal molecules or other biological active chemical substances are an important entity class in life science publications and patents. Several representations and nomenclatures for chemicals like SMILES, InChI, IUPAC or trivial names exist. Only SMILES and InChI names allow a direct structure search, but in biomedical texts trivial names and Iupac like names are used more frequent. While trivial names can be found with a dictionary-based approach and in such a way mapped to their corresponding structures, it is not possible to enumerate all IUPAC names. In this work, we present a new machine learning approach based on conditional random fields (CRF) to find mentions of IUPAC and IUPAC-like names in scientific text as well as its evaluation and the conversion rate with available name-to-structure tools. Results: We present an IUPAC name recognizer with an F₁ measure of 85.6% on a MEDLINE corpus. The evaluation of different CRF orders and offset conjunction orders demonstrates the importance of these parameters. An evaluation of hand-selected patent sections containing large enumerations and terms with mixed nomenclature shows a good performance on these cases (F₁ measure 81.5%). Remaining recognition problems are to detect correct borders of the typically long terms, especially when occurring in parentheses or enumerations. We demonstrate the scalability of our implementation by providing results from a full MEDLINE run. Availability: We plan to publish the corpora, annotation guideline as well as the conditional random field model as a UIMA component. Contact: roman.klinger@scai.fraunhofer.de