MetImp is a web tool for -omics missing data imputation, especially for mass spectrometry-based metabolomics data from metabolic profiling and targeted analysis.
There are three types of missing values in Metabolomics: missing completely at random (MCAR), missing at random (MAR), and missing not at random (MNAR). We recommend a comprehensive approach for handling missing values:
(1) Analyze the possible reasons of missing values: whether they are MNAR/MAR, or MNAR;
(2) If necessary, check raw data and/or adjust parameter settings of data preprocessing in order to fill back certain missing values in an accurate way;
(3) Apply missing filtering to remove those unreliable variables with unacceptable number of missing values;
(4) If missing value imputation is needed for your data, choose an appropriate method;
(5) MetImp provides:
A. Group-wise missing filtering: 0%~100% adjustable, 80% is the default proportion ;
B. For MCAR/MAR: we provide RF, SVD, kNN, Mean, Median imputation, and RF is default;
C. For MNAR: we provide GSimp, QRILC, HM, Zero, Binary imputation, and GSimp is default.For the imputation evaluation pipeline, please visit: https://github.com/WandeRum/MVI-evaluation
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License
For sending comments, suggestions, bug reports of MetImp, please email to Runmin Wei (firstname.lastname@example.org)
Stekhoven, Daniel J., and Peter Bühlmann. "MissForest—non-parametric missing value imputation for mixed-type data." Bioinformatics 28.1 (2011): 112-118.
Hastie, Trevor, et al. "Imputing missing data for gene expression arrays." (1999): 1-7.
Troyanskaya, Olga, et al. "Missing value estimation methods for DNA microarrays." Bioinformatics 17.6 (2001): 520-525.
Lazar, Cosmin. "imputeLCMD: a collection of methods for left-censored missing data imputation." R package, version 2 (2015).