Search

Ribosomes are information-processing macromolecular machines that integrate complex sequence patterns in messenger RNA (mRNA) transcripts to synthesize proteins. Studies of the sequence features that distinguish mRNAs from long noncoding RNAs (lncRNAs) may yield insight into the information that directs and regulates translation. Computational methods for calculating protein-coding potential are important for distinguishing mRNAs from lncRNAs during genome annotation, but most machine learning methods for this task rely on previously known rules to define features. Sequence-to-sequence (seq2seq) models, particularly ones using transformer networks, have proven capable of learning complex grammatical relationships between words to perform natural language translation. Seeking to leverage these advancements in the biological domain, we present a seq2seq formulation for predicting protein-coding potential with deep neural networks and demonstrate that simultaneously learning translation from RNA to protein improves classification performance relative to a classification-only training objective. Inspired by classical signal processing methods for gene discovery and Fourier-based image-processing neural networks, we introduce LocalFilterNet (LFNet). LFNet is a network architecture with an inductive bias for modeling the three-nucleotide periodicity apparent in coding sequences. We incorporate LFNet within an encoder-decoder framework to test whether the translation task improves the classification of transcripts and the interpretation of their sequence features. We use the resulting model to compute nucleotide-resolution importance scores, revealing sequence patterns that could assist the cellular machinery in distinguishing mRNAs and lncRNAs. Finally, we develop a novel approach for estimating mutation effects from Integrated Gradients, a backpropagation-based feature attribution, and characterize the difficulty of efficient approximations in this setting., Competing Interests: The authors have declared that no competing interests exist., (Copyright: © 2023 Valencia, Hendrix. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.) more...

Ribosomes are information-processing macromolecular machines that integrate complex sequence patterns in messenger RNA (mRNA) transcripts to synthesize proteins. Studies of the sequence features that distinguish mRNAs from long noncoding RNAs (lncRNAs) may yield insight into the information that directs and regulates translation. Computational methods for calculating protein-coding potential are important for distinguishing mRNAs from lncRNAs during genome annotation, but most machine learning methods for this task rely on previously known rules to define features. Sequence-to-sequence (seq2seq) models, particularly ones using transformer networks, have proven capable of learning complex grammatical relationships between words to perform natural language translation. Seeking to leverage these advancements in the biological domain, we present a seq2seq formulation for predicting protein-coding potential with deep neural networks and demonstrate that simultaneously learning translation from RNA to protein improves classification performance relative to a classification-only training objective. Inspired by classical signal processing methods for gene discovery and Fourier-based image-processing neural networks, we introduce LocalFilterNet (LFNet). LFNet is a network architecture with an inductive bias for modeling the three-nucleotide periodicity apparent in coding sequences. We incorporate LFNet within an encoder-decoder framework to test whether the translation task improves the classification of transcripts and the interpretation of their sequence features. We use the resulting model to compute nucleotide-resolution importance scores, revealing sequence patterns that could assist the cellular machinery in distinguishing mRNAs and lncRNAs. Finally, we develop a novel approach for estimating mutation effects from Integrated Gradients, a backpropagation-based feature attribution, and characterize the difficulty of efficient approximations in this setting., Competing Interests: 6Competing Interest Statement The authors have no competing interests to declare. more...

Class I Myosins are a subfamily of motor proteins with ATPase activity and a characteristic structure conserved in all myosins: A N-Terminal Motor Domain, a central Neck and a C terminal Tail domain. Humans have eight genes for these myosins. Class I Myosins have different functions: regulate membrane tension, participate in endocytosis, exocytosis, intracellular trafficking and cell migration. Cell migration is influenced by many cellular components including motor proteins, like myosins. Recently has been reported that changes in myosin expression have an impact on the migration of cancer cells, the formation of infiltrates and metastasis. We propose that class I myosins might be potential markers for future diagnostic, prognostic or even as therapeutic targets in leukemia and other cancers. Abbreviations: Myo1g: Myosin 1g; ALL: Acute Lymphoblastic Leukemia, TH1: Tail Homology 1; TH2: Tail Homology 2; TH3: Tail Homology 3. more...

Objective: To describe the current nationwide perspectives and practice regarding intraoperative oxygen titration in cardiac surgery., Design: Prospective, observational survey., Setting: Hospitals across the United States., Participants: Cardiovascular anesthesiologists and perfusionists., Interventions: Expert- and consensus-derived electronic surveys were sent to perfusionists and cardiac anesthesiologists to evaluate the current intraoperative practices around oxygen administration. Providers were asked about individual intraoperative oxygen titration practices used at different stages of cardiac surgical procedures. Anonymous responses were collected in the Research Electronic Data Capture (REDCap)., Measurements and Main Results: A total of 3,335 providers were invited to participate, of whom 554 (317 anesthesiologists and 237 perfusionists) were included in the final analysis (17% response rate). During cardiopulmonary bypass (CPB), perfusionists reported a median (interquartile range [IQR]) target range from 150 (110-220)-to-325 mmHg (250-400), while anesthesiologists reported a significantly lower target range from 90 (70-150)-to-250 mmHg (158-400) (p values <0.0001 and 0.02, respectively). This difference was most pronounced at lower partial pressure of arterial oxygen (PaO 2 ) ranges. The median PaO 2 considered "too low" by perfusionists was 100 mmHg (IQR 80-125), whereas it was 60 mmHg (IQR 60-75) for anesthesiologists, who reported for both off and on bypass. The median PaO 2 considered "too high" was 375 mmHg (IQR 300-400) for perfusionists and 300 mmHg (IQR 200-400) for anesthesiologists. Anesthesiologists, therefore, reported more comfort with significantly lower PaO 2 values (p < 0.0001), and considered a higher PaO 2 value less desirable compared with perfusionists (p < 0.0001)., Conclusions: This survey demonstrated there was wide variation in oxygen administration practices between perfusionists and anesthesiologists. Hyperoxygenation was more common while on CPB., (Copyright © 2022. Published by Elsevier Inc.) more...

Preterm neonates are at high risk of infectious and inflammatory diseases which require antibiotic treatment. Antibiotics influence neonatal gut microbiome development, and intestinal dysbiosis has been associated with delayed gastrointestinal transit. Neonates who take less time to pass meconium have a better tolerance to enteral feeding. We analyzed the effect of neonatal antibiotic treatment on the stool pattern and oral tolerance in 106 preterm infants < 33 weeks gestational age. Neonates were classified in 3 groups according to neonatal antibiotic (ABT) treatment days: no antibiotics, 3−7 d ABT, and ≥8 d ABT. Preterm infants from the ≥8 d ABT group took longer to pass meconium and to start green and yellow stools, took longer to reach 100 and 150 mL/kg/day, and reached reduced volumes in enteral feeds at day of life 14 and 28 than infants from no ABT and 3−7 d ABT groups. Multiple linear regression models showed that neonatal antibiotic treatment, birth weight, invasive mechanical ventilation, surfactant, enteral feeding start day, neonatal parenteral nutrition, and neonatal fasting days are associated with the stool pattern and oral tolerance in preterm infants. more...