scholarly journals Stability of feature selection utilizing Graph Convolutional Neural Network and Layer-wise Relevance Propagation

2021 ◽  
Author(s):  
Hryhorii Chereda ◽  
Andreas Leha ◽  
Tim Beissbarth
Keyword(s):  
Neural Network ◽  
Gene Expression ◽  
Feature Selection ◽  
Deep Learning ◽  
Patient Specific ◽  
Molecular Signatures ◽  
Expression Data ◽  
Selection Methods ◽  
Feature Sets ◽  
The Stability

Motivation: High-throughput technologies play a more and more significant role in discovering prognostic molecular signatures and identifying novel drug targets. It is common to apply Machine Learning (ML) methods to classify high-dimensional gene expression data and to determine a subset of features (genes) that is important for decisions of a ML model. One feature subset of important genes corresponds to one dataset and it is essential to sustain the stability of feature sets across different datasets with the same clinical endpoint since the selected genes are candidates for prognostic biomarkers. The stability of feature selection can be improved by including information of molecular networks into ML methods. Gene expression data can be assigned to the vertices of a molecular network's graph and then classified by a Graph Convolutional Neural Network (GCNN). GCNN is a contemporary deep learning approach that can be applied to graph-structured data. Layer-wise Relevance Propagation (LRP) is a technique to explain decisions of deep learning methods. In our recent work we developed Graph Layer-wise Relevance Propagation (GLRP) --- a method that adapts LRP to a graph convolution and explains patient-specific decisions of GCNN. GLRP delivers individual molecular signatures as patient-specific subnetworks that are parts of a molecular network representing background knowledge about biological mechanisms. GLRP gives a possibility to deliver the subset of features corresponding to a dataset as well, so that the stability of feature selection performed by GLRP can be measured and compared to that of other methods. Results: Utilizing two large breast cancer datasets, we analysed properties of feature sets selected by GLRP (GCNN+LRP) such as stability and permutation importance. We have implemented a graph convolutional layer of GCNN as a Keras layer so that the SHAP (SHapley Additive exPlanation) explanation method could be also applied to a Keras version of a GCNN model. We compare the stability of feature selection performed by GCNN+LRP to the stability of GCNN+SHAP and to other ML based feature selection methods. We conclude, that GCNN+LRP shows the highest stability among other feature selection methods including GCNN+SHAP. It was established that the permutation importance of features among GLRP subnetworks is lower than among GCNN+SHAP subnetworks, but in the context of the utilized molecular network, a GLRP subnetwork of an individual patient is on average substantially more connected (and interpretable) than a GCNN+SHAP subnetwork, which consists mainly of single vertices.

scholarly journals Inference of Genetic Networks From Time-Series and Static Gene Expression Data: Combining a Random-Forest-Based Inference Method With Feature Selection Methods

2020 ◽  
Vol 11 ◽  
Author(s):  
Shuhei Kimura ◽  
Ryo Fukutomi ◽  
Masato Tokuhisa ◽  
Mariko Okada
Keyword(s):  
Gene Expression ◽  
Feature Selection ◽  
Random Forest ◽  
Gene Expression Data ◽  
Computational Cost ◽  
Expression Data ◽  
Selection Methods ◽  
Inference Method ◽  
Combined Application ◽  
Inference Methods

Several researchers have focused on random-forest-based inference methods because of their excellent performance. Some of these inference methods also have a useful ability to analyze both time-series and static gene expression data. However, they are only of use in ranking all of the candidate regulations by assigning them confidence values. None have been capable of detecting the regulations that actually affect a gene of interest. In this study, we propose a method to remove unpromising candidate regulations by combining the random-forest-based inference method with a series of feature selection methods. In addition to detecting unpromising regulations, our proposed method uses outputs from the feature selection methods to adjust the confidence values of all of the candidate regulations that have been computed by the random-forest-based inference method. Numerical experiments showed that the combined application with the feature selection methods improved the performance of the random-forest-based inference method on 99 of the 100 trials performed on the artificial problems. However, the improvement tends to be small, since our combined method succeeded in removing only 19% of the candidate regulations at most. The combined application with the feature selection methods moreover makes the computational cost higher. While a bigger improvement at a lower computational cost would be ideal, we see no impediments to our investigation, given that our aim is to extract as much useful information as possible from a limited amount of gene expression data.

scholarly journals Incorporating Pathway Information into Feature Selection towards Better Performed Gene Signatures

2019 ◽  
Vol 2019 ◽  
pp. 1-12 ◽  
Author(s):  
Suyan Tian ◽  
Chi Wang ◽  
Bing Wang
Keyword(s):  
Gene Expression ◽  
Feature Selection ◽  
Gene Expression Data ◽  
Gene Selection ◽  
Selection Process ◽  
Biological Knowledge ◽  
Expression Data ◽  
Selection Methods ◽  
Its Gene ◽  
Active Research

To analyze gene expression data with sophisticated grouping structures and to extract hidden patterns from such data, feature selection is of critical importance. It is well known that genes do not function in isolation but rather work together within various metabolic, regulatory, and signaling pathways. If the biological knowledge contained within these pathways is taken into account, the resulting method is a pathway-based algorithm. Studies have demonstrated that a pathway-based method usually outperforms its gene-based counterpart in which no biological knowledge is considered. In this article, a pathway-based feature selection is firstly divided into three major categories, namely, pathway-level selection, bilevel selection, and pathway-guided gene selection. With bilevel selection methods being regarded as a special case of pathway-guided gene selection process, we discuss pathway-guided gene selection methods in detail and the importance of penalization in such methods. Last, we point out the potential utilizations of pathway-guided gene selection in one active research avenue, namely, to analyze longitudinal gene expression data. We believe this article provides valuable insights for computational biologists and biostatisticians so that they can make biology more computable.

CLASSIFYING TEMPORAL MICROARRAY DATA BY SELECTING INFORMATIVE GENES

2013 ◽  
Vol 11 (03) ◽  
pp. 1341006
Author(s):  
QIANG LOU ◽  
ZORAN OBRADOVIC
Keyword(s):  
Gene Expression ◽  
Feature Selection ◽  
Gene Expression Data ◽  
Microarray Data ◽  
Data Sets ◽  
Temporal Data ◽  
Expression Data ◽  
Selection Methods ◽  
Temporal Gene Expression ◽  
Single Matrix

In order to more accurately predict an individual's health status, in clinical applications it is often important to perform analysis of high-dimensional gene expression data that varies with time. A major challenge in predicting from such temporal microarray data is that the number of biomarkers used as features is typically much larger than the number of labeled subjects. One way to address this challenge is to perform feature selection as a preprocessing step and then apply a classification method on selected features. However, traditional feature selection methods cannot handle multivariate temporal data without applying techniques that flatten temporal data into a single matrix in advance. In this study, a feature selection filter that can directly select informative features from temporal gene expression data is proposed. In our approach, we measure the distance between multivariate temporal data from two subjects. Based on this distance, we define the objective function of temporal margin based feature selection to maximize each subject's temporal margin in its own relevant subspace. The experimental results on synthetic and two real flu data sets provide evidence that our method outperforms the alternatives, which flatten the temporal data in advance.

scholarly journals A convolutional neural network for predicting transcriptional regulators of genes in Arabidopsis transcriptome data reveals classification based on positive regulatory interactions

2019 ◽  
Author(s):  
Dan MacLean
Keyword(s):  
Neural Network ◽  
Gene Expression ◽  
Neural Networks ◽  
Deep Learning ◽  
Convolutional Neural Network ◽  
Convolutional Neural Networks ◽  
Expression Profiles ◽  
Biological Data ◽  
Expression Data ◽  
Data Set

AbstractGene Regulatory networks that control gene expression are widely studied yet the interactions that make them up are difficult to predict from high throughput data. Deep Learning methods such as convolutional neural networks can perform surprisingly good classifications on a variety of data types and the matrix-like gene expression profiles would seem to be ideal input data for deep learning approaches. In this short study I compiled training sets of expression data using the Arabidopsis AtGenExpress global stress expression data set and known transcription factor-target interactions from the Arabidopsis PLACE database. I built and optimised convolutional neural networks with a best model providing 95 % accuracy of classification on a held-out validation set. Investigation of the activations within this model revealed that classification was based on positive correlation of expression profiles in short sections. This result shows that a convolutional neural network can be used to make classifications and reveal the basis of those calssifications for gene expression data sets, indicating that a convolutional neural network is a useful and interpretable tool for exploratory classification of biological data. The final model is available for download and as a web application.

scholarly journals Explaining decisions of Graph Convolutional Neural Networks: patient-specific molecular subnetworks responsible for metastasis prediction in breast cancer

2020 ◽  
Author(s):  
Hryhorii Chereda ◽  
Annalen Bleckmann ◽  
Kerstin Menck ◽  
Júlia Perera-Bel ◽  
Philip Stegmaier ◽  
...  
Keyword(s):  
Breast Cancer ◽  
Gene Expression ◽  
Neural Networks ◽  
Deep Learning ◽  
Convolutional Neural Networks ◽  
Gene Expression Data ◽  
Molecular Network ◽  
Patient Specific ◽  
Expression Data ◽  
Learning Methods

AbstractMotivationContemporary deep learning approaches show cutting-edge performance in a variety of complex prediction tasks. Nonetheless, the application of deep learning in healthcare remains limited since deep learning methods are often considered as non-interpretable black-box models. Layer-wise Relevance Propagation (LRP) is a technique to explain decisions of deep learning methods. It is widely used to interpret Convolutional Neural Networks (CNNs) applied on image data. Recently, CNNs started to extend towards non-euclidean domains like graphs. Molecular networks are commonly represented as graphs detailing interactions between molecules. Gene expression data can be assigned to the vertices of these graphs. In other words, gene expression data can be structured by utilizing molecular network information as prior knowledge. Graph-CNNs can be applied to structured gene expression data, for example, to predict metastatic events in breast cancer. Therefore, there is a need for explanations showing which part of a molecular network is relevant for predicting an event, e.g. distant metastasis in cancer, for each individual patient.ResultsWe extended the procedure of LRP to make it available for Graph-CNN and tested its applicability on a large breast cancer dataset. We present Graph Layer-wise Relevance Propagation (GLRP) as a new method to explain the decisions made by Graph-CNNs. We demonstrate a sanity check of the developed GLRP on a hand-written digits dataset, and then applied the method on gene expression data. We show that GLRP provides patient-specific molecular subnetworks that largely agree with clinical knowledge and identify common as well as novel, and potentially druggable, drivers of tumor progression. As a result this method could be potentially highly useful on interpreting classification results on the individual patient level, as for example in precision medicine approaches or a molecular tumor board.Availabilityhttps://gitlab.gwdg.de/UKEBpublic/graph-lrphttps://frankkramer-lab.github.io/MetaRelSubNetVis/[email protected]

Sign in / Sign up

Export Citation Format

Share Document