Deep learning reveals 3D atherosclerotic plaque distribution and composition.

Complications of atherosclerosis are the leading cause of morbidity and mortality worldwide. Various genetically modified mouse models are used to investigate disease trajectory with classical histology, currently the preferred methodology to elucidate plaque composition. Here, we show the strength of light-sheet fluorescence microscopy combined with deep learning image analysis for characterising and quantifying plaque burden and composition in whole aorta specimens. 3D imaging is a non-destructive method that requires minimal ex vivo handling and can be up-scaled to large sample sizes. Combined with deep learning, atherosclerotic plaque in mice can be identified without any ex vivo staining due to the autofluorescent nature of the tissue. The aorta and its branches can subsequently be segmented to determine how anatomical position affects plaque composition and progression. Here, we find the highest plaque accumulation in the aortic arch and brachiocephalic artery. Simultaneously, aortas can be stained for markers of interest (for example the pan immune cell marker CD45) and quantified. In ApoE-/- mice we observe that levels of CD45 reach a plateau after which increases in plaque volume no longer correlate to immune cell infiltration. All underlying code is made publicly available to ease adaption of the method.

Quantitative whole-brain 3D imaging of tyrosine hydroxylase-labeled neuron architecture in the mouse MPTP model of Parkinson's disease.

Parkinson's disease (PD) is a basal ganglia movement disorder characterized by progressive degeneration of the nigrostriatal dopaminergic system. Immunohistochemical methods have been widely used for characterization of dopaminergic neuronal injury in animal models of PD, including the MPTP (1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine) mouse model. However, conventional immunohistochemical techniques applied to tissue sections have inherent limitations with respect to loss of 3D resolution, yielding insufficient information on the architecture of the dopaminergic system. To provide a more comprehensive and non-biased map of MPTP-induced changes in central dopaminergic pathways, we used iDISCO immunolabeling, light-sheet fluorescence microscopy (LSFM) and deep-learning computational methods for whole-brain three-dimensional visualization and automated quantitation of tyrosine hydroxylase (TH)-positive neurons in the adult mouse brain. Mice terminated 7 days after acute MPTP administration demonstrated widespread alterations in TH expression. Compared to vehicle controls, MPTP-dosed mice showed a significant loss of TH-positive neurons in the substantia nigra pars compacta and ventral tegmental area. Also, MPTP dosing reduced overall TH signal intensity in basal ganglia nuclei, i.e. the substantia nigra, caudate-putamen, globus pallidus and subthalamic nucleus. In contrast, increased TH signal intensity was predominantly observed in limbic regions, including several subdivisions of the amygdala and hypothalamus. In conclusion, mouse whole-brain 3D imaging is ideal for unbiased automated counting and densitometric analysis of TH-positive cells. The LSFM-deep learning pipeline tracked brain-wide changes in catecholaminergic pathways in the MPTP mouse model of PD, and may be applied for preclinical characterization of compounds targeting dopaminergic neurotransmission.

NetSurfP-2.0: Improved prediction of protein structural features by integrated deep learning.

The ability to predict local structural features of a protein from the primary sequence is of paramount importance for unraveling its function in absence of experimental structural information. Two main factors affect the utility of potential prediction tools: their accuracy must enable extraction of reliable structural information on the proteins of interest, and their runtime must be low to keep pace with sequencing data being generated at a constantly increasing speed. Here, we present NetSurfP-2.0, a novel tool that can predict the most important local structural features with unprecedented accuracy and runtime. NetSurfP-2.0 is sequence-based and uses an architecture composed of convolutional and long short-term memory neural networks trained on solved protein structures. Using a single integrated model, NetSurfP-2.0 predicts solvent accessibility, secondary structure, structural disorder, and backbone dihedral angles for each residue of the input sequences. We assessed the accuracy of NetSurfP-2.0 on several independent test datasets and found it to consistently produce state-of-the-art predictions for each of its output features. We observe a correlation of 80% between predictions and experimental data for solvent accessibility, and a precision of 85% on secondary structure 3-class predictions. In addition to improved accuracy, the processing time has been optimized to allow predicting more than 1000 proteins in less than 2 hours, and complete proteomes in less than 1 day.

Computational Methods for Identification of T Cell Neoepitopes in Tumors.

Cancer immunotherapy has experienced several major breakthroughs in the past decade. Most recently, technical advances in next-generation sequencing methods have enabled discovery of tumor-specific mutations leading to protective T cell neoepitopes. Many of the successes are enabled by computational methods, which facilitate processing of raw data, mapping of mutations, and prediction of neoepitopes. In this book chapter, we provide an overview of the computational tasks related to the identification of neoepitopes, propose specific tools and best practices, and discuss strengths, weaknesses, and future challenges.

An introduction to deep learning on biological sequence data: examples and solutions.

MOTIVATION: Deep neural network architectures such as convolutional and long short-term memory networks have become increasingly popular as machine learning tools during the recent years. The availability of greater computational resources, more data, new algorithms for training deep models and easy to use libraries for implementation and training of neural networks are the drivers of this development. The use of deep learning has been especially successful in image recognition; and the development of tools, applications and code examples are in most cases centered within this field rather than within biology. RESULTS: Here, we aim to further the development of deep learning methods within biology by providing application examples and ready to apply and adapt code templates. Given such examples, we illustrate how architectures consisting of convolutional and long short-term memory neural networks can relatively easily be designed and trained to state-of-the-art performance on three biological sequence problems: prediction of subcellular localization, protein secondary structure and the binding of peptides to MHC Class II molecules. AVAILABILITY AND IMPLEMENTATION: All implementations and datasets are available online to the scientific community at https://github.com/vanessajurtz/lasagne4bio. CONTACT: skaaesonderby@gmail.com. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MetaPhinder-Identifying Bacteriophage Sequences in Metagenomic Data Sets.

Bacteriophages are the most abundant biological entity on the planet, but at the same time do not account for much of the genetic material isolated from most environments due to their small genome sizes. They also show great genetic diversity and mosaic genomes making it challenging to analyze and understand them. Here we present MetaPhinder, a method to identify assembled genomic fragments (i.e.contigs) of phage origin in metagenomic data sets. The method is based on a comparison to a database of whole genome bacteriophage sequences, integrating hits to multiple genomes to accomodate for the mosaic genome structure of many bacteriophages. The method is demonstrated to out-perform both BLAST methods based on single hits and methods based on k-mer comparisons. MetaPhinder is available as a web service at the Center for Genomic Epidemiology https://cge.cbs.dtu.dk/services/MetaPhinder/, while the source code can be downloaded from https://bitbucket.org/genomicepidemiology/metaphinder or https://github.com/vanessajurtz/MetaPhinder.

HostPhinder: A Phage Host Prediction Tool.

The current dramatic increase of antibiotic resistant bacteria has revitalised the interest in bacteriophages as alternative antibacterial treatment. Meanwhile, the development of bioinformatics methods for analysing genomic data places high-throughput approaches for phage characterization within reach. Here, we present HostPhinder, a tool aimed at predicting the bacterial host of phages by examining the phage genome sequence. Using a reference database of 2196 phages with known hosts, HostPhinder predicts the host species of a query phage as the host of the most genomically similar reference phages. As a measure of genomic similarity the number of co-occurring k-mers (DNA sequences of length k) is used. Using an independent evaluation set, HostPhinder was able to correctly predict host genus and species for 81% and 74% of the phages respectively, giving predictions for more phages than BLAST and significantly outperforming BLAST on phages for which both had predictions. HostPhinder predictions on phage draft genomes from the INTESTI phage cocktail corresponded well with the advertised targets of the cocktail. Our study indicates that for most phages genomic similarity correlates well with related bacterial hosts. HostPhinder is available as an interactive web service [1] and as a stand alone download from the Docker registry [2].