Introduction | Comparison of the performances of the method to popular feature selection and classification algorithms shows that or strategy is effective in identifying microbial clades associated to the different sample groups, providing a novel analysis method for targeted metagenomic datasets. |
Predictive classification pipeline | We compared the predictive performance of PhyloRelief with the Random Forest classifier (PhyloRelief +RF) to LEfSe + RF, MetaPhyl (without feature selection) and Random Forest used as both classifier and feature selection method (RF + RF). |
Predictive classification pipeline | In order to avoid overfitting and selection bias effects, the feature selection procedure was included in the cross validation loop [40,41]. |
Predictive classification pipeline | In the case of LEfSe + RF, LEfSe was treated as feature selection method using the common p-value threshold of 0.05. |
Predictivity of the ranked features in supervised classification problems | The Random Forest (RF) classifier was recently proven to be the most effective in this class of problems [26,27] , both for feature selection and classification. |
Predictivity of the ranked features in supervised classification problems | We compared the performance of PhyloRelief coupled With the RF classifier to LEfSe [30], an algorithm that uses statistical tests for biomarker discovery, to MetaPhyl, a recent phylogeny-based method for the classification of microbial communities [31] and to Random Forest, used both as classifier and feature selection method. |
Discussion | To account for redundancy, we have used representative, common approaches including feature selection within the learning algorithm (via regularization), feature filtering (via feature clustering), and feature combination (via principal components analysis). |
Supervised learning: Classification | Furthermore, in order to assess the effect of reducing redundancy and focusing on the most interpretable feature contributions, three different sets of input features were considered: the complete set (20 features: 4 subclasses * 5 antigens), the filtered set with one feature selected from each cluster based on correlation with function (6 features), and the PC features (7 leading PCs), as illustrated in Fig 2. |
Supervised learning: Classification | Though the goal of this study was not to comprehensively and rigorously assess feature selection methods, which would require further subsampling the data, we did investigate the sensitivity of the cluster-based filtering to our use of the features within each cluster that had the highest PCC. |
Supervised learning: Classification | The PC features selected for the cytokines are more consistent with the other feature sets, with PC2 (IgG2/4 vs. 1/3) modulated by PC6 (V1V2), along with an IgG4.V1V2 down-selection via PC7. |
Supervised learning: Regression | While the disappointing performance of the more sophisticated methods could potentially be improved by custom feature selection methods or parameter tuning, our goal here is not to provide such a benchmark but rather to establish the general scheme of predictive modeling of antibody feature: function relationships. |
Excitatory and inhibitory STDP cooperatively shape structured lateral connections | From a linear analysis, we can expect that When gY1 is positive, E-to-I connections tend to be feature selective (see Eq (35) in Methods). |
Excitatory and inhibitory STDP cooperatively shape structured lateral connections | We can evaluate feature selectivity of inhibitory neurons by where QYA and QYB are the sets of excitatory neurons responding preferentially to sources A and B, respectively. |
Excitatory and inhibitory STDP cooperatively shape structured lateral connections | Indeed, when the LTD time window is narrow, analytically calculated ng tends to take negative values (the green line in Fig 6A), and E-to-I connections organized in the simulation are not feature selective (the blue points in Fig 6A). |
P P ji) : vSGf(W;§)Zqiuqlu7 : VSG§(WJ§)ZquqZfl7 “=1 M=1 < 1-,.) 2[ dsF<w;§,s>[ drax<r>[ dt'¢<t'>¢<t' — (r — s + 2%)), —oo 0 max(0,r—s+2dXd) < 1,) 2[ dsF<w;§,s>[0 dram 0 dqay<q>[0 dr'am | In this approximation, we additionally assume that w and the eigenvalue is es lY — Because the eigenvector develops by eXp[ch1Y — Wfift], When ng is positive, the E-to-I connections are more likely to be structured in a way that the inhibitory neurons become feature selective . |
STDP in E-to-I and I-to-E connections | In our model, although inhibitory neurons are not directly projected from input sources, as excitatory neurons learn a specific input source (Fig 5D, left panel), inhibitory neurons acquire feature selectivity through Hebbian STDP at synaptic connections from those excitatory neurons (Fig 5D, middle panel). |
Discussion | We thus used feature selection to derive a small set of genes which capture the prognostic power of signalling entropy independently of other clinical variables, thus representing a more readily applicable quantifier of stem-ness and intra-tumour heterogeneity. |
Discussion | In comparing signalling entropy to signatures such as MammaPrint it is worth pointing out that a direct comparison is unfair signalling entropy does not involve feature selection . |
Discussion | Although signalling entropy was not found to outperform existing prognostic markers in lung adenocarcinoma, by using the SE score, derived by signalling entropy guided feature selection , it was possible to outperform existing state of the art prognostic factors such as CADMI eXpression across independent data sets. |
Signalling entropy’s prognostic power in breast cancer can be represented by a small number of genes | By using signalling entropy to refine a set of prognostic genes identified by Cox regression, our approach refines the feature selection approach based on correlation with outcome [24]. |
Signalling entropy’s prognostic power in breast cancer can be represented by a small number of genes | Criticism of feature selection for prognostic classifiers based on gene sets ranked by correlation with outcome has stemmed from the considerable discordance of such features between data sets [47, 48]. |
Bright-Field Head Identification | Second, in the feature selection step, distinct mathematical descriptors that may help to describe and distinguish the structure of interest are calculated for each layer of classification. |
Bright-Field Head Identification | However, in addition to informative feature selection and the curation of a representative training set, the performance of SVM classification models is subject to several parameters associated with the model itself and its kernel function [34, 48]. |
Discussion | Finally, while utility of our framework will require feature selection and training for each particular application, the modularity and architecture of our framework permits aspects of the specific tools we have developed here to be reused. |
Feature selection by mutual information | Feature selection by mutual information |
Feature selection by mutual information | Mutual information is a stochastic measure of dependence [69] and it has been widely applied in feature selection in order to find an informative subset for model training [70]. |
Feature selection by mutual information | All the analyses in this study other than the cross-validation of model used the features selected from the complete data. |