Multiple Imputation Approaches Applied to the Missing Value Problem in Bottom-Up Proteomics.

Additionally, there was a negative trend toward non-significance with the increasing number of missing values in all methods, excluding the SFI-hybrid ( Figure S2 ).

also in 4,201 other papers

This observation suggests that either the methods have a bias to choose complete cases, or the algorithms are imputing values too close to the observed to be considered significant; The statistics with single MAR or MNAR strategies (not the SFI-hybrid) are negatively impacted by increasing number and type of missingness, characterized by large standard deviations, logFC sign fluctuations and an overall trend toward non-significance as seen by the loss in the number of significant proteins from the ground truth and known protein complex interactors; To avoid unnecessarily excluding data as in a complete case analysis, a combinatorial MAR/MNAR approach, such as SFI-hybrid, that imputes missing values sep

also in 4,871 other papers