Pirat

Pirat imputes missing values (either MNARs or MCARs) in bottom-up LC-MS/MS proteomics data using a penalized maximum likelihood strategy that models instrument censorship and accounts for sibling peptide correlations.

Key Features:

• Penalized maximum likelihood imputation: Imputes missing values using a penalized maximum likelihood strategy without requiring parameter tuning.
• Instrument censorship modeling: Models the instrument limit by learning a global censoring mechanism from the available data.
• Sibling peptide correlation: Accounts for correlations between enzymatic cleavage products (peptides or precursor ions) by estimating their covariance matrix.
• Transcriptomics integration: Leverages complementary transcriptomics measurements when multi-omic assays are available.

Scientific Applications:

• Proteomics missing value imputation: Handles complex and heterogeneous missing values in label-free bottom-up proteomics using mass spectrometry and liquid chromatography.
• Differential analysis support: Improves downstream differential analysis by providing more accurate imputation compared to pre-existing methods.
• Multi-omic data integration: Integrates transcriptomic information to enhance imputation performance in multi-omic experimental designs.

Methodology:

Pirat uses penalized likelihood optimization for censored missing value imputation, learning a global censoring mechanism from available data and estimating the covariance matrix between enzymatic cleavage products.