Biology subjects

Lio, P.

Publications and source records attributed to Lio, P..

49 records · Page 3Linked to original sources

bioRxiv2019

Volumetric Segmentation and Characterisation of the Paracingulate Sulcus on MRI Scans

Many architectures of deep neural networks have been designed to solve specific biomedical problems, among which segmentation is a critical step to detect and locate the boundaries of the target object from an image. In this paper, we develop a deep learning based framework to automatically segment the paracingulate sulcus (PCS) from the MRI scan and estimate lengths for its segments. The study is the first work on segmentation and characterisation of the PCS, and the model achieves a Dice score of over 0.77 on segmentation, which demonstrates its potential for clinical use to assist human annotation. Moreover, the proposed architecture as a solution can be generalised to other problems where the object has similar patterns.

bioinformatics↗

bioRxiv2019

Unsupervised Machine Learning for Data Encoding appliedto Ovarian Cancer Transcriptomes

Machine learning algorithms are revolutionising how information can be extracted from complex and high-dimensional data sets via intelligent compression. For example, unsupervised Autoen-coders train a deep neural network with a low-dimensional "bottlenecked" central layer to reconstruct input vectors. Variational Autoencoders (VAEs) have shown promise at learning meaningful latent spaces for text, image and more recently, gene-expression data. In the latter case they have been shown capable of capturing biologically relevant features such as a patients sex or tumour type. Here we train a VAE on ovarian cancer transcriptomes from The Cancer Genome Atlas and show that, in many cases, the latent spaces learns an encoding predictive of cisplatin chemotherapy resistance. We analyse the effectiveness of such an architecture to a wide range of hyperparameters as well as use a state-of-the-art clustering algorithm, t-SNE, to embed the data in a two-dimensional manifold and visualise the predictive power of the trained latent spaces. By correlating genes to resistance-predictive encodings we are able to extract biological processes likely responsible for platinum resistance. Finally we demonstrate that variational autoencoders can reliably encode gene expression data contaminated with significant amounts of Gaussian and dropout noise, a necessary feature if this technique is to be applicable to other data sets, including those in non-medical fields.

cancer biology↗

bioRxiv2019

Unsupervised generative and graph representation learning for modelling cell differentiation

Using machine learning techniques to build representations from biomedical data can help us understand the latent biological mechanism of action and lead to important discoveries. Recent developments in single-cell RNA-sequencing protocols have allowed measuring gene expression for individual cells in a population, thus opening up the possibility of finding answers to biomedical questions about cell differentiation. In this paper, we explore unsupervised generative neural methods, based on the variational autoencoder, that can model cell differentiation by building meaningful representations from the high dimensional and complex gene expression data. We use disentanglement methods based on information theory to improve the data representation and achieve better separation of the biological factors of variation in the gene expression data. In addition, we use a graph autoencoder consisting of graph convolutional layers to predict relationships between single-cells. Based on these models, we develop a computational framework that consists of methods for identifying the cell types in the dataset, finding driver genes for the differentiation process and obtaining a better understanding of relationships between cells. We illustrate our methods on datasets from multiple species and also from different sequencing technologies.

bioinformatics↗

bioRxiv2019

scAEspy: a unifying tool based on autoencoders for the analysis of single-cell RNA sequencing data

BackgroundSingle-cell RNA sequencing (scRNA-Seq) experiments are gaining ground to study the molecular processes that drive normal development as well as the onset of different pathologies. Finding an effective and efficient low-dimensional representation of the data is one of the most important steps in the downstream analysis of scRNA-Seq data, as it could provide a better identification of known or putatively novel cell-types. Another step that still poses a challenge is the integration of different scRNA-Seq datasets. Though standard computational pipelines to gain knowledge from scRNA-Seq data exist, a further improvement could be achieved by means of machine learning approaches. ResultsAutoencoders (AEs) have been effectively used to capture the non-linearities among gene interactions of scRNA-Seq data, so that the deployment of AE-based tools might represent the way forward in this context. We introduce here scAEspy, a unifying tool that embodies: (1) four of the most advanced AEs, (2) two novel AEs that we developed on purpose, (3) different loss functions. We show that scAEspy can be coupled with various batch-effect removal tools to integrate data by different scRNA-Seq platforms, in order to better identify the cell-types. We benchmarked scAEspy against the most used batch-effect removal tools, showing that our AE-based strategies outperform the existing solutions. ConclusionsscAEspy is a user-friendly tool that enables using the most recent and promising AEs to analyse scRNA-Seq data by only setting up two user-defined parameters. Thanks to its modularity, scAEspy can be easily extended to accommodate new AEs to further improve the downstream analysis of scRNA-Seq data. Considering the relevant results we achieved, scAEspy can be considered as a starting point to build a more comprehensive toolkit designed to integrate multi single-cell omics.

bioinformatics↗

bioRxiv2019

A machine learning tool for interpreting differences in cognition using brain features

Predicting variability in cognition traits is an attractive and challenging area of research, where different approaches and datasets have been implemented with mixed results. Some powerful Machine Learning algorithms employed before are difficult to interpret, while other algorithms are easy to interpret but might not be as powerful. To improve understanding of individual cognitive differences in humans, we make use of the most recent developments in Machine Learning in which powerful prediction models can be interpreted with confidence. We used neuroimaging data and a variety of behavioural, cognitive, affective and health measures from 905 people obtained from the Human Connectome Project (HCP). As a main contribution of this paper, we show how one could interpret the neuroanatomical basis of cognition, with recent methods which we believe are not yet fully explored in the field. By reducing neuroimages to a well characterised set of features generated from surface-based morphometry and cortical myelin estimates, we make the interpretation of such models easier as each feature is self-explanatory. The code used in this tool is available in a public repository: https://github.com/tjiagoM/interpreting-cognition-paper-2019

neuroscience↗

bioRxiv2018

Genetic Effects of Welding Fumes on the progression ofNeurodegenerative Diseases

BackgroundWelding exposes different types of fumes, gases and radiant energy that can be potentially dangerous for unsafe welders health. Welding fumes (WFs) are a significant problem among all those exposed. WFs are a complex mixture of metallic oxides, silicates and fluorides that may result in different health effects. If a welder inhales such fumes in large quantities over a long period of time, there is a risk of various neurodegenerative diseases (NDGDs) development.\n\nMethodsWe developed quantitative frameworks to identify the genetic relationship of WFs and NDGDs. We analyzed Gene Expression microarray data from WFs exposed tissues and NDGDs including Parkinsons disease (PD), Alzheimers disease (AD), Lou Gehrigs disease (LGD), Epilepsy disease (ED), Multiple Sclerosis disease (MSD) datasets. We constructed disease-gene relationship networks and identified dysregulated pathways, ontological path- ways and protein-protein interaction sub-network using multilayer network topology and neighborhood-based benchmarking.\n\nResultsWe observed that WFs shares 18, 16, 13, 19 and 19 differentially expressed genes with PD, AD, LGD, ED and MSD respectively. Gene expression dysregulation along with relationship networks, pathways and ontologic analysis showed that WFs are responsible for the progression of PD, AD, LGD, ED and MSD neurodegenerative diseases.\n\nConclusionOur developed network-based approach to analysis and investigate the genetic effects of welding fumes on PD, AD, LGD, ED and MSD neurodegenerative diseases could be helpful to understand the causal influences of WF exposure for the progression of the NDGDs.

bioinformatics↗

bioRxiv2018

Genetic Effects of Welding Fumes to the development ofRespiratory System Diseases

BackgroundThe welding process releases potentially hazardous gases and fumes, mainly composed of metallic oxides, fluorides and silicates. Long term welding fume (WF) inhalation is a recognized health issue that carries a risk of developing chronic health problems, particularly respiratory system diseases (RSDs). Aside from general airway irritation, WFs may drive direct cellular responses in the respiratory system which increase risk of RSD, but these are not well understood.\n\nMethodsWe developed a quantitative framework to identify gene expression effects of WFs that may affect RSD development. We analyzed gene expression microarray data from WF-exposed tissues and RSD-affected tissues, including chronic bronchitis (CB), asthma (AS), pulmonary edema (PE), lung cancer (LC) datasets. We built disease-gene (disea-some) association networks and identified dysregulated signaling and ontological pathways, and protein-protein interaction sub-network using neighborhood-based benchmarking and multilayer network topology.\n\nResultsWe observed many genes with altered expression in WF-exposed tissues were also among differentially expressed genes (DEGs) in RSD tissues; for CB, AS, PE and LC there were 34, 27, 50 and 26 genes respectively. DEGs analysis, with disease association networks, pathways, ontological analysis and protein-protein interaction sub-network suggest significant links between WF exposure and the development of CB, AS, PE and LC.\n\nConclusionsOur network-based analysis and investigation of the genetic links of WFs and RSDs confirm a number of genes and gene products are plausible participants in RSD development. Our results are a significant resource to identify causal influences on the development of RSDs, particularly in the context of WF exposure.

bioinformatics↗

bioRxiv2018

Genetic effect of type 2 Diabetes to the progression of Neurological Diseases

Neurological Diseases (NDs) are progressive disorder often advances with age and comorbidities of Type 2 diabetes (T2D). Epidemiological, clinical and neuropathological evidence advocate that patients with T2D are at an increased risk of getting NDs. However, it is very little known how T2D affects the risk and severity of NDs.\n\nTo tackle these problems, we employed a transcriptional analysis of affected tissues using agnostic approaches to identify overlapping cellular functions. In this study, we examined gene expression microarray human datasets along with control and disease-affected individuals. Differentially expressed genes (DEG) were identified for both T2D and NDs that includes Alzheimer Disease (AD), Parkinson Disease (PD), Amyotrophic Lateral Sclerosis (ALS), Epilepsy Disease (ED), Huntington Disease (HD), Cerebral Palsy (CP) and Multiple Sclerosis Disease (MSD).\n\nWe have developed genetic association and diseasome network of T2D and NDs based on the neighborhood-based benchmarking and multilayer network topology approaches. Overlapping DEG sets go through protein-protein interaction for hub protein identification and gene enrichment using pathway analysis and gene ontology methods that enhance our understanding of the fundamental molecular procedure of NDs progression.\n\nGene expression analysis platforms have been extensively used to investigate altered pathways and to identify potential biomarkers and drug targets. Finally, we validated our identified biomarkers using the gold benchmark datasets which identified the corresponding relationship of T2D and NDs. Therapeutic targets aimed at attenuating identified altered pathway could ameliorate neurological dysfunction in a T2D patient.

bioinformatics↗

bioRxiv2018

A parameter-efficient deep learning approach to predict conversion from mild cognitive impairment to Alzheimer’s disease within three years

Some forms of mild cognitive impairment (MCI) can be the clinical precursor of severe dementia like Alzheimers disease (AD), while other types of MCI tend to remain stable over-time and do not progress to AD pathology. To choose an effective and personalized treatment for AD, we need to identify which MCI patients are at risk of developing AD and which are not.\n\nHere, we present a novel deep learning architecture, based on dual learning and an ad hoc layer for 3D separable convolutions, which aims at identifying those people with MCI who have a high likelihood of developing AD. Our deep learning procedures combine structural magnetic resonance imaging (MRI), demographic, neuropsychological, and APOe4 genotyping data as input measures. The most novel characteristics of our machine learning model compared to previous ones are as follows: 1) multi-tasking, in the sense that our deep learning model jointly learns to simultaneously predict both MCI to AD conversion, and AD vs healthy classification which facilitates the relevant feature extraction for prognostication; 2) the neural network classifier employs relatively few parameters compared to other deep learning architectures (we use ~550,000 network parameters, orders of magnitude lower than other network designs) without compromising network complexity and hence significantly limits data-overfitting; 3) both structural MRI images and warp field characteristics, which quantify the amount of volumetric change compared to the common template, were used as separate input streams to extract as much information as possible from the MRI data. All the analyses were performed on a subset of the Alzheimers Disease Neuroimaging Initiative (ADNI) database, for a total of n=785 participants (192 AD, 409 MCI, and184 healthy controls (HC)).\n\nWe found that the most predictive combination of inputs included the structural MRI images and the demographic, neuropsychological, and APOe4 data, while the warp field metric added little predictive value. We achieved an area under the ROC curve (AUC) of 0.925 with a 10-fold cross-validated accuracy of 86%, a sensitivity of 87.5% and specificity of 85% in classifying MCI patients who developed AD in three years time from those individuals showing stable MCI over the same time-period. To the best of our knowledge, this is the highest performance reported on a test set achieved in the literature using similar data. The same network provided an AUC of 1 and 100% accuracy, sensitivity and specificity when classifying NC from AD. We also demonstrated that our classification framework was robust to different co-registration templates and possibly irrelevant features / image sections.\n\nOur approach is flexible and can in principle integrate other imaging modalities, such as PET, and a more diverse group of clinical data. The convolutional framework is potentially applicable to any 3D image dataset and gives the flexibility to design a computer-aided diagnosis system targeting the prediction of any medical condition utilizing multi-modal imaging and tabular clinical data.

bioinformatics↗

bioRxiv2018

GEESE: Metabolically driven latent space learning for gene expression data

Gene expression microarrays provide a characterisation of the transcriptional activity of a particular biological sample. Their high dimensionality hampers the process of pattern recognition and extraction. Several approaches have been proposed for gleaning information about the hidden structure of the data. Among these approaches, deep generative models provide a powerful way for approximating the manifold on which the data reside.\n\nHere we develop GEESE, a deep learning based framework that provides novel insight into the manifold learning for gene expression data, employing a metabolic model to constrain the learned representation. We evaluated the proposed framework, showing its ability to capture biologically relevant features, and encoding that features in a much simpler latent space. We showed how using a metabolic model to drive the autoencoder learning process helps in achieving better generalisation to unseen data. GEESE provides a novel perspective on the problem of unsupervised learning for biological data.\n\nAvailabilitySource code of GEESE is available at https://bitbucket.org/mbarsacchi/geese/.

systems biology↗

bioRxiv2018

Network-Based Biomarkers Enable Cross-Disease Biomarker Discovery

Biomarkers lie at the heart of precision medicine, biodiversity monitoring, agricultural pathogen detection, amongst others. Surprisingly, while rapid genomic profiling is becoming ubiquitous, the development of biomarkers almost always involves the application of bespoke techniques that cannot be directly applied to other datasets. There is an urgent need for a systematic methodology to create biologically-interpretable molecular models that robustly predict key phenotypes. We therefore created SIMMS: an algorithm that fragments pathways into functional modules and uses these to predict phenotypes. We applied SIMMS to multiple data-types across four diseases, and in each it reproducibly identified subtypes, made superior predictions to the best bespoke approaches, and identified known and novel signaling nodes. To demonstrate its ability on a new dataset, we measured 33 genes/nodes of the PIK3CA pathway in 1,734 FFPE breast tumours and created a four-subnetwork prediction model. This model significantly out-performed existing clinically-used molecular tests in an independent 1,742-patient validation cohort. SIMMS is generic and can work with any molecular data or biological network, and is freely available at: https://cran.r-project.org/web/packages/SIMMS.

bioinformatics↗

bioRxiv2017

Paratope Prediction using Convolutional and Recurrent Neural Networks

Antibodies play an essential role in the immune system of vertebrates and are vital tools in research and diagnostics. While hypervariable regions of antibodies, which are responsible for binding, can be readily identified from their amino acid sequence, it remains challenging to accurately pinpoint which amino acids will be in contact with the antigen (the paratope). In this work, we present a sequence-based probabilistic machine learning algorithm for paratope prediction, named Parapred. Parapred uses a deep-learning architecture to leverage features from both local residue neighbourhoods and across the entire sequence. The method outperforms the current state-of-the-art methodology, and only requires a stretch of amino acid sequence corresponding to a hypervariable region as an input, without any information about the antigen. We further show that our predictions can be used to improve both speed and accuracy of a rigid docking algorithm. The Parapred method is freely available at https://github.com/eliberis/parapred for download.

bioinformatics↗

bioRxiv2017

Single-cell RNA-Sequencing uncovers transcriptional states and fate decisionsin haematopoiesis

The success of marker-based approaches for dissecting haematopoiesis in mouse and human is reliant on the presence of well-defined cell-surface markers specific for diverse progenitor populations. An inherent problem with this approach is that the presence of specific cell surface markers does not directly reflect the transcriptional state of a cell. Here we used a marker-free approach to computationally reconstruct the blood lineage tree in zebrafish and order cells along their differentiation trajectory, based on their global transcriptional differences. Within the population of transcriptionally similar stem and progenitor cells our analysis revealed considerable cell-to-cell differences in their probability to transition to another, committed state. Once fate decision was executed, the suppression of transcription of ribosomal genes and up-regulation of lineage specific factors coordinately controlled lineage differentiation. Evolutionary analysis further demonstrated that this haematopoietic program was highly conserved between zebrafish and higher vertebrates.

cell biology↗