A Review of Imputation Strategies for Isobaric Labeling-Based Shotgun Proteomics

Abstract

The throughput efficiency and increased depth of coverage provided by isobaric-labeled proteomics measurements have led to increased usage of these techniques. However, the structure of missing data is different than unlabeled studies, which prompts the need for this review to compare the efficacy of nine imputation methods on large isobaric-labeled proteomics data sets to guide researchers on the appropriateness of various imputation methods. Imputation methods were evaluated by accuracy, statistical hypothesis test inference, and run time. In general, expectation maximization and random forest imputation methods yielded the best performance, and constant-based methods consistently performed poorly across all data set sizes and percentages of missing values. For data sets with small sample sizes and higher percentages of missing data, results indicate that statistical inference with no imputation may be preferable. On the basis of the findings in this review, there are core imputation methods that perform better for isobaric-labeled proteomics data, but great care and consideration as to whether imputation is the optimal strategy should be given for data sets comprised of a small number of samples.

Replication of single-cell proteomics data reveals important computational challenges

“…Although most proteins are common to several MS runs, each run exhibits a specific set of proteins that were probably present but missed in the other runs (Figure 3B). We observe only marginal correlation between the peptide abundance and the proportion of missingness in single-cell samples (Figure 3C), as was already described for bulk proteomics [31]. While upcoming technical improvements to SCP will further decrease the amount of missing values, computational approaches will still be required.…”

Can we put Humpty Dumpty back together again? What does protein quantification mean in bottom-up proteomics?

“…Evaluating large, multiplexed experiments in comparison to label-free approaches, the pattern of the missing data appears to be very distinct, but the macrostructure overall is similar in regard to the relationship between abundance and missing data (8). These multiplexing methods are still limited in the number of samples that can be uniquely tagged, combined, and analyzed at once, and while multiple batches of samples can be acquired, the same peptides are less frequently sampled in different batches compared to proteins (9).…”

DreamAI: algorithm for the imputation of proteomics data

“…al. (2014) [22] proposed an exponential function based probability model to characterize the abundance dependent missing pattern in unlabeled LC-MS/MS proteomics data, and a penalized EM algorithm (PEMM) to fit the model, which can be used for imputation[23]. To further accommodate the multiplex level missing structure in the labeled proteomics data, the authors further introduced mixEMM [10], which not only models the abundance dependent batch-level missing pattern but also utilizes mixed effects to better account for experimental variations across multiplexes.…”

“…But the investigation is incomprehensive, due to the limited number of imputation methods considered and the inadequate numerical examples with rather simplified missing mechanism assumptions. In another recent review work [23], Bramer et. al.…”