ILRC: a hybrid biomarker discovery algorithm based on improved L1 regularization and clustering in microarray data

Abstract

Background Finding significant genes or proteins from gene chip data for disease diagnosis and drug development is an important task. However, the challenge comes from the curse of the data dimension. It is of great significance to use machine learning methods to find important features from the data and build an accurate classification model. Results The proposed method has proved superior to the published advanced hybrid feature selection method and traditional feature selection method on different public microarray data sets. In addition, the biomarkers selected using our method show a match to those provided by the cooperative hospital in a set of clinical cleft lip and palate data. Method In this paper, a feature selection algorithm ILRC based on clustering and improved L1 regularization is proposed. The features are firstly clustered, and the redundant features in the sub-clusters are deleted. Then all the remaining features are iteratively evaluated using ILR. The final result is given according to the cumulative weight reordering. Conclusion The proposed method can effectively remove redundant features. The algorithm’s output has high stability and classification accuracy, which can potentially select potential biomarkers.

The Prognostic Value and Immune Landscapes of a m6A/m5C/m1A-Related LncRNAs Signature in Head and Neck Squamous Cell Carcinoma

“…L1-regularization greatly increases the stability of modeling and achieves the purpose of selecting characteristic variables. The most outstanding advantage of Lasso lies in the penalized regression analysis of all variable coefficients, the relatively unimportant independent variable coefficients become 0, which is excluded from modeling ( Vaid et al, 2021 ; Yu et al, 2021 ; Hoq et al, 2021 ). However, the 44 lncRNAs are too large for constructing a prognosis model, so we used LASSO regression analysis to further compress the 44 lncRNAs to get the lncRNAs with the most prognostic value, and their correlation coefficients.…”

The Effect of Missing Data Mechanisms on Deep Learning in Binary Classification: A Simulation Study

“…In terms of regularization, correlation structure is important along with the optimization of the regularization parameters. 25 Naqvi et al conducted a simulation study on the classification of clinical mastitis in cows with and without missing values in the input variables. 26 They concluded that deep learning performance is still good in the scenarios with missing values.…”

BAMBI: Integrativebiostatistical andartificial-intelligencemodels discover coding and non-coding RNA genes asbiomarkers

“…We compared the performance of BAMBI and current alternative methods (BioDISCML (Leclercq et al, 2019), ILRC (Yu et al, 2021), ECMarker (Jin et al, 2021)) in two RNA-seq datasets (breast cancer dataset and psoriasis dataset) and two microarray datasets (colon cancer dataset (Alon et al, 1999) and prostate cancer dataset (Singh et al, 2002). Because the split of train and test data will strong influence the performance score of the target method, especially for small datasets with less than 1000 samples, we applied a cross validation comparison structure to obtain a convincing comparison result for different methods.…”

“…Several methods have been proposed for RNA biomarker identification. We compared BAMBI with the other state-of-the-art methods coupled with computational tools: BioDiscML (Leclercq et al, 2019), ILRC (Yu et al, 2021), ECMarker (Jin et al, 2021). Because all the other methods cannot process RNA-seq data directly and cannot normalize microarray data by default, we used the RNA gene expression table generated by Phase II of the BAMBI as the input files for the other methods.…”