rMisbeta: A Robust Missing Value Imputation Approach in Transcriptomics and Metabolomics Data

AbstractLongitudinal datasets of human ageing studies usually have a high volume of missing data, and one way to handle missing values in a dataset is to replace them with estimations. However, there are many methods to estimate missing values, and no single method is the best for all datasets. In this article, we propose a data-driven missing value imputation approach that performs a feature-wise selection of the best imputation method, using known information in the dataset to rank the five methods we selected, based on their estimation error rates. We evaluated the proposed approach in two sets of experiments: a classifier-independent scenario, where we compared the applicabilities and error rates of each imputation method; and a classifier-dependent scenario, where we compared the predictive accuracy of Random Forest classifiers generated with datasets prepared using each imputation method and a baseline approach of doing no imputation (letting the classification algorithm handle the missing values internally). Based on our results from both sets of experiments, we concluded that the proposed data-driven missing value imputation approach generally resulted in models with more accurate estimations for missing data and better performing classifiers, in longitudinal datasets of human ageing. We also observed that imputation methods devised specifically for longitudinal data had very accurate estimations. This reinforces the idea that using the temporal information intrinsic to longitudinal data is a worthwhile endeavour for machine learning applications, and that can be achieved through the proposed data-driven approach.

Download Full-text

GSimp: A Gibbs sampler based left-censored missing value imputation approach for metabolomics studies

PLoS Computational Biology ◽

10.1371/journal.pcbi.1005973 ◽

2018 ◽

Vol 14 (1) ◽

pp. e1005973 ◽

Cited By ~ 25

Author(s):

Runmin Wei ◽

Jingye Wang ◽

Erik Jia ◽

Tianlu Chen ◽

Yan Ni ◽

...

Keyword(s):

Gibbs Sampler ◽

Missing Value ◽

Missing Value Imputation ◽

Imputation Approach

Download Full-text

Missing Value Imputation Approach for Mass Spectrometry-based Metabolomics Data

10.1101/171967 ◽

2017 ◽

Cited By ~ 1

Author(s):

Runmin Wei ◽

Jingye Wang ◽

Mingming Su ◽

Erik Jia ◽

Tianlu Chen ◽

...

Keyword(s):

Mass Spectrometry ◽

Missing Values ◽

Pearson Correlation ◽

Imputation Accuracy ◽

Metabolomics Data ◽

Missing Value ◽

Sample Distribution ◽

Imputation Methods ◽

Missing Value Imputation ◽

Squared Error

AbstractIntroductionMissing values exist widely in mass-spectrometry (MS) based metabolomics data. Various methods have been applied for handling missing values, but the selection of methods can significantly affect following data analyses and interpretations. According to the definition, there are three types of missing values, missing completely at random (MCAR), missing at random (MAR), and missing not at random (MNAR).ObjectivesThe aim of this study was to comprehensively compare common imputation methods for different types of missing values using two separate metabolomics data sets (977 and 198 serum samples respectively) to propose a strategy to deal with missing values in metabolomics studies.MethodsImputation methods included zero, half minimum (HM), mean, median, random forest (RF), singular value decomposition (SVD), k-nearest neighbors (kNN), and quantile regression imputation of left-censored data (QRILC). Normalized root mean squared error (NRMSE) and NRMSE-based sum of ranks (SOR) were applied to evaluate the imputation accuracy for MCAR/MAR and MNAR correspondingly. Principal component analysis (PCA)/partial least squares (PLS)-Procrustes sum of squared error were used to evaluate the overall sample distribution. Student’s t-test followed by Pearson correlation analysis was conducted to evaluate the effect of imputation on univariate statistical analysis.ResultsOur findings demonstrated that RF imputation performed the best for MCAR/MAR and QRILC was the favored one for MNAR.ConclusionCombining with “modified 80% rule”, we proposed a comprehensive strategy and developed a public-accessible web-tool for missing value imputation in metabolomics data.

Download Full-text

Missing value imputation for LC-MS metabolomics data by incorporating metabolic network and adduct ion relations

Bioinformatics ◽

10.1093/bioinformatics/btx816 ◽

2017 ◽

Vol 34 (9) ◽

pp. 1555-1561 ◽

Cited By ~ 7

Author(s):

Zhuxuan Jin ◽

Jian Kang ◽

Tianwei Yu

Keyword(s):

Metabolic Network ◽

Metabolomics Data ◽

Missing Value ◽

Missing Value Imputation ◽

Ion Relations

Download Full-text

Missing value imputation strategies for metabolomics data

Electrophoresis ◽

10.1002/elps.201500352 ◽

2015 ◽

Vol 36 (24) ◽

pp. 3050-3060 ◽

Cited By ~ 55

Author(s):

Emily Grace Armitage ◽

Joanna Godzien ◽

Vanesa Alonso-Herranz ◽

Ángeles López-Gonzálvez ◽

Coral Barbas

Keyword(s):

Metabolomics Data ◽

Missing Value ◽

Missing Value Imputation

Download Full-text

APPLICATION OF AN EMPIRICAL GROWTH MODEL AND MULTIPLE IMPUTATION IN HARD DISK DRIVE FIELD RETURN PREDICTION

International Journal of Reliability Quality and Safety Engineering ◽

10.1142/s0218539310003950 ◽

2010 ◽

Vol 17 (06) ◽

pp. 565-577 ◽

Cited By ~ 6

Author(s):

SHAOANG ZHANG ◽

FENG-BIN SUN ◽

ROSS GOUGH

Keyword(s):

Hard Disk Drive ◽

Growth Model ◽

Survival Data ◽

Early Stage ◽

Hard Disk ◽

Disk Drive ◽

Model Estimation ◽

Missing Value ◽

Missing Value Imputation ◽

Imputation Approach

Appropriate model assumption and robust model estimation are critical for accurate hard disk drive (HDD) field return prediction. Parametric distribution models that are conventionally used in HDD reliability analyses do not describe well HDD field survival. A biological growth model, Bertalanfy-Richards growth model, is introduced in this paper to model HDD field survival within warranty period. A nonparametric multiple missing value imputation approach is used to improve model estimation when not enough observations of field returns are available at early stage of product life. Tests of the model and the missing value imputation approach using actual HDD field survival data suggest they perform well in describing and predicting HDD warranty field returns.

Download Full-text

GSimp: A Gibbs sampler based left-censored missing value imputation approach for metabolomics studies

10.1101/177410 ◽

2017 ◽

Author(s):

Runmin Wei ◽

Jingye Wang ◽

Erik Jia ◽

Tianlu Chen ◽

Yan Ni ◽

...

Keyword(s):

Gibbs Sampler ◽

Real World ◽

Missing Values ◽

Statistical Analyses ◽

Targeted Metabolomics ◽

Missing Not At Random ◽

Missing Value ◽

Imputation Methods ◽

Missing Value Imputation ◽

Imputation Approach

AbstractLeft-censored missing values commonly exist in targeted metabolomics datasets and can be considered as missing not at random (MNAR). Improper data processing procedures for missing values will cause adverse impacts on subsequent statistical analyses. However, few imputation methods have been developed and applied to the situation of MNAR in the field of metabolomics. Thus, a practical left-censored missing value imputation method is urgently needed. We have developed an iterative Gibbs sampler based left-censored missing value imputation approach (GSimp). We compared GSimp with other three imputation methods on two real-world targeted metabolomics datasets and one simulation dataset using our imputation evaluation pipeline. The results show that GSimp outperforms other imputation methods in terms of imputation accuracy, observation distribution, univariate and multivariate analyses, and statistical sensitivity. The R code for GSimp, evaluation pipeline, vignette, real-world and simulated targeted metabolomics datasets are available at: https://github.com/WandeRum/GSimp.Author summaryMissing values caused by the limit of detection/quantification (LOD/LOQ) were widely observed in mass spectrometry (MS)-based targeted metabolomics studies and could be recognized as missing not at random (MNAR). MNAR leads to biased parameter estimations and jeopardizes following statistical analyses in different aspects, such as distorting sample distribution, impairing statistical power, etc. Although a wide range of missing value imputation methods was developed for –omics studies, a limited number of methods was designed appropriately for the situation of MNAR currently. To alleviate problems caused by MNAR and facilitate targeted metabolomics studies, we developed a Gibbs sampler based missing value imputation approach, called GSimp, which is public-accessible on GitHub. And we compared our method with existing approaches using an imputation evaluation pipeline on real-world and simulated metabolomics datasets to demonstrate the superiority of our method from different perspectives.

Download Full-text

A New Approach of Outlier-robust Missing Value Imputation for Metabolomics Data Analysis

Current Bioinformatics ◽

10.2174/1574893612666171121154655 ◽

2018 ◽

Vol 14 (1) ◽

pp. 43-52 ◽

Cited By ~ 3

Author(s):

Nishith Kumar ◽

Md. Aminul Hoque ◽

Md. Shahjaman ◽

S.M. Shahinul Islam ◽

Md. Nurul Haque Mollah

Keyword(s):

Mass Spectrometry ◽

Data Analysis ◽

Missing Values ◽

Real Data ◽

Gas Chromatography Mass Spectrometry ◽

Data Generation ◽

Metabolomics Data ◽

Missing Value ◽

Missing Value Imputation ◽

Chromatography Mass Spectrometry

Background: Metabolomics data generation and quantification are different from other types of molecular “omics” data in bioinformatics. Mass spectrometry (MS) based (gas chromatography mass spectrometry (GC-MS), liquid chromatography mass spectrometry (LC-MS), etc.) metabolomics data frequently contain missing values that make some quantitative analysis complex. Typically metabolomics datasets contain 10% to 20% missing values that originate from several reasons, like analytical, computational as well as biological hazard. Imputation of missing values is a very important and interesting issue for further metabolomics data analysis. </P><P> Objective: This paper introduces a new algorithm for missing value imputation in the presence of outliers for metabolomics data analysis. </P><P> Method: Currently, the most well known missing value imputation techniques in metabolomics data are knearest neighbours (kNN), random forest (RF) and zero imputation. However, these techniques are sensitive to outliers. In this paper, we have proposed an outlier robust missing imputation technique by minimizing twoway empirical mean absolute error (MAE) loss function for imputing missing values in metabolomics data. Results: We have investigated the performance of the proposed missing value imputation technique in a comparison of the other traditional imputation techniques using both simulated and real data analysis in the absence and presence of outliers. Conclusion: Results of both simulated and real data analyses show that the proposed outlier robust missing imputation technique is better performer than the traditional missing imputation methods in both absence and presence of outliers.

Download Full-text

Metabolomic Biomarker Identification in Presence of Outliers and Missing Values

BioMed Research International ◽

10.1155/2017/2437608 ◽

2017 ◽

Vol 2017 ◽

pp. 1-11 ◽

Cited By ~ 9

Author(s):

Nishith Kumar ◽

Md. Aminul Hoque ◽

Md. Shahjaman ◽

S. M. Shahinul Islam ◽

Md. Nurul Haque Mollah

Keyword(s):

Data Analysis ◽

High Throughput ◽

Missing Values ◽

Real Data ◽

Data Matrix ◽

Metabolomics Data ◽

Missing Value ◽

Biomarker Identification ◽

Missing Value Imputation ◽

High Throughput Technology

Metabolomics is the sophisticated and high-throughput technology based on the entire set of metabolites which is known as the connector between genotypes and phenotypes. For any phenotypic changes, potential metabolite (biomarker) identification is very important because it provides diagnostic as well as prognostic markers and can help to develop new biomolecular therapy. Biomarker identification from metabolomics data analysis is hampered by the use of high-throughput technology that provides high dimensional data matrix which contains missing values as well as outliers. However, missing value imputation and outliers handling techniques play important role in identifying biomarker correctly. Although several missing value imputation techniques are available, outliers deteriorate the accuracy of imputation as well as the accuracy of biomarker identification. Therefore, in this paper we have proposed a new biomarker identification technique combining the groupwise robust singular value decomposition, t-test, and fold-change approach that can identify biomarkers more correctly from metabolomics dataset. We have also compared the performance of the proposed technique with those of other traditional techniques for biomarker identification using both simulated and real data analysis in absence and presence of outliers. Using our proposed method in hepatocellular carcinoma (HCC) dataset, we have also identified the four upregulated and two downregulated metabolites as potential metabolomic biomarkers for HCC disease.

Download Full-text