Biological data annotation via a human-augmenting AI-based labeling system.
NPJ digital medicine
2021; 4 (1): 145
Biology has become a prime area for the deployment of deep learning and artificial intelligence (AI), enabled largely by the massive data sets that the field can generate. Key to most AI tasks is the availability of a sufficiently large, labeled data set with which to train AI models. In the context of microscopy, it is easy to generate image data sets containing millions of cells and structures. However, it is challenging to obtain large-scale high-quality annotations for AI models. Here, we present HALS (Human-Augmenting Labeling System), a human-in-the-loop data labeling AI, which begins uninitialized and learns annotations from a human, in real-time. Using a multi-part AI composed of three deep learning models, HALS learns from just a few examples and immediately decreases the workload of the annotator, while increasing the quality of their annotations. Using a highly repetitive use-case-annotating cell types-and running experiments with seven pathologists-experts at the microscopic analysis of biological specimens-we demonstrate a manual work reduction of 90.60%, and an average data-quality boost of 4.34%, measured across four use-cases and two tissue stain types.
View details for DOI 10.1038/s41746-021-00520-6
View details for PubMedID 34620993
TDP43 ribonucleoprotein granules: physiologic function to pathologic aggregates.
Ribonucleoprotein (RNP) assemblies are ubiquitous in eukaryotic cells and have functions throughout RNA transcription, splicing, and stability. Of the RNA-binding proteins that form RNPs, TAR DNA-binding protein of 43 kD (TDP43) is of particular interest due to its essential nature and its association with disease. TDP43 plays critical roles in RNA metabolism, many of which require its recruitment to RNP granules such as stress granules, myo-granules, and neuronal transport granules. Moreover, the presence of cytoplasmic TDP43-positive inclusions is a pathological hallmark of several neurodegenerative diseases. Despite the pervasiveness of TDP43 aggregates, TDP43 mutations are exceedingly rare, suggesting that aggregation may be linked to dysregulation of TDP43 function. Oligomerization is a part of normal TDP43 function; thus, it is of interest to understand what triggers the irreversible aggregation that is seen in disease. Herein, we examine TDP43 functions, particularly in RNP granules, and the mechanisms which may explain pathological TDP43 aggregation.
View details for DOI 10.1080/15476286.2021.1963099
View details for PubMedID 34412568
Tau aggregates are RNA-protein assemblies that mislocalize multiple nuclear speckle components.
Tau aggregates contribute to neurodegenerative diseases, including frontotemporal dementia and Alzheimer's disease (AD). Although RNA promotes tau aggregation invitro, whether tau aggregates in cells contain RNA is unknown. We demonstrate, in cell culture and mouse brains, that cytosolic and nuclear tau aggregates contain RNA with enrichment for small nuclear RNAs (snRNAs) and small nucleolar RNAs (snoRNAs). Nuclear tau aggregates colocalize with and alter the composition, dynamics, and organization of nuclear speckles, membraneless organelles involved in pre-mRNA splicing. Moreover, several nuclear speckle components, including SRRM2, mislocalize to cytosolic tau aggregates in cells, mouse brains, and brains of individuals with AD, frontotemporal dementia (FTD), and corticobasal degeneration (CBD). Consistent with these alterations, we observe that the presence of tau aggregates is sufficient to alter pre-mRNA splicing. This work identifies tau alteration of nuclear speckles as a feature of tau aggregation that may contribute to the pathology of tau aggregates.
View details for DOI 10.1016/j.neuron.2021.03.026
View details for PubMedID 33848474