Dataset Information

Machine learning for cluster analysis of localization microscopy data.

ABSTRACT: Quantifying the extent to which points are clustered in single-molecule localization microscopy data is vital to understanding the spatial relationships between molecules in the underlying sample. Many existing computational approaches are limited in their ability to process large-scale data sets, to deal effectively with sample heterogeneity, or require subjective user-defined analysis parameters. Here, we develop a supervised machine-learning approach to cluster analysis which is fast and accurate. Trained on a variety of simulated clustered data, the neural network can classify millions of points from a typical single-molecule localization microscopy data set, with the potential to include additional classifiers to describe different subtypes of clusters. The output can be further refined for the measurement of cluster area, shape, and point-density. We demonstrate this approach on simulated data and experimental data of the kinase Csk and the adaptor PAG in primary human T cell immunological synapses.

SUBMITTER: Williamson DJ

PROVIDER: S-EPMC7083906 | biostudies-literature | 2020 Mar

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Machine learning for cluster analysis of localization microscopy data.

Williamson David J DJ Burn Garth L GL Simoncelli Sabrina S Griffié Juliette J Peters Ruby R Davis Daniel M DM Owen Dylan M DM

Nature communications 20200320 1

Quantifying the extent to which points are clustered in single-molecule localization microscopy data is vital to understanding the spatial relationships between molecules in the underlying sample. Many existing computational approaches are limited in their ability to process large-scale data sets, to deal effectively with sample heterogeneity, or require subjective user-defined analysis parameters. Here, we develop a supervised machine-learning approach to cluster analysis which is fast and accu ...[more]

PMID: 32198352

Similar Datasets

Project description:Caveolae are plasma membrane invaginations whose formation requires caveolin-1 (Cav1), the adaptor protein polymerase I, and the transcript release factor (PTRF or CAVIN1). Caveolae have an important role in cell functioning, signaling, and disease. In the absence of CAVIN1/PTRF, Cav1 forms non-caveolar membrane domains called scaffolds. In this work, we train machine learning models to automatically distinguish between caveolae and scaffolds from single molecule localization microscopy (SMLM) data. We apply machine learning algorithms to discriminate biological structures from SMLM data. Our work is the first that is leveraging machine learning approaches (including deep learning models) to automatically identifying biological structures from SMLM data. In particular, we develop and compare three binary classification methods to identify whether or not a given 3D cluster of Cav1 proteins is a caveolae. The first uses a random forest classifier applied to 28 hand-crafted/designed features, the second uses a convolutional neural net (CNN) applied to a projection of the point clouds onto three planes, and the third uses a PointNet model, a recent development that can directly take point clouds as its input. We validate our methods on a dataset of super-resolution microscopy images of PC3 prostate cancer cells labeled for Cav1. Specifically, we have images from two cell populations: 10 PC3 and 10 CAVIN1/PTRF-transfected PC3 cells (PC3-PTRF cells) that form caveolae. We obtained a balanced set of 1714 different cellular structures. Our results show that both the random forest on hand-designed features and the deep learning approach achieve high accuracy in distinguishing the intrinsic features of the caveolae and non-caveolae biological structures. More specifically, both random forest and deep CNN classifiers achieve classification accuracy reaching 94% on our test set, while the PointNet model only reached 83% accuracy. We also discuss the pros and cons of the different approaches.

Project description:Fungal secondary metabolites (SMs) contribute to the diversity of fungal ecological communities, niches, and lifestyles. Many fungal SMs have one or more medically and industrially important activities (e.g., antifungal, antibacterial, and antitumor). The genes necessary for fungal SM biosynthesis are typically located right next to each other in the genome and are known as biosynthetic gene clusters (BGCs). However, whether fungal SM bioactivity can be predicted from specific attributes of genes in BGCs remains an open question. We adapted machine learning models that predicted SM bioactivity from bacterial BGC data with accuracies as high as 80% to fungal BGC data. We trained our models to predict the antibacterial, antifungal, and cytotoxic/antitumor bioactivity of fungal SMs on two data sets: (i) fungal BGCs (data set comprised of 314 BGCs) and (ii) fungal (314 BGCs) and bacterial BGCs (1,003 BGCs). We found that models trained on fungal BGCs had balanced accuracies between 51% and 68%, whereas training on bacterial and fungal BGCs had balanced accuracies between 56% and 68%. The low prediction accuracy of fungal SM bioactivities likely stems from the small size of the data set; this lack of data, coupled with our finding that including bacterial BGC data in the training data did not substantially change accuracies currently limits the application of machine learning approaches to fungal SM studies. With >15,000 characterized fungal SMs, millions of putative BGCs in fungal genomes, and increased demand for novel drugs, efforts that systematically link fungal SM bioactivity to BGCs are urgently needed.IMPORTANCEFungi are key sources of natural products and iconic drugs, including penicillin and statins. DNA sequencing has revealed that there are likely millions of biosynthetic pathways in fungal genomes, but the chemical structures and bioactivities of >99% of natural products produced by these pathways remain unknown. We used artificial intelligence to predict the bioactivities of diverse fungal biosynthetic pathways. We found that the accuracies of our predictions were generally low, between 51% and 68%, likely because the natural products and bioactivities of only very few fungal pathways are known. With >15,000 characterized fungal natural products, millions of putative biosynthetic pathways present in fungal genomes, and increased demand for novel drugs, our study suggests that there is an urgent need for efforts that systematically identify fungal biosynthetic pathways, their natural products, and their bioactivities.

Dataset Information

Machine learning for cluster analysis of localization microscopy data.

Publications

Machine learning for cluster analysis of localization microscopy data.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets