MAP: model-based analysis of proteomic data to detect proteins with significant abundance changes

Abstract

Isotope-labeling-based mass spectrometry (MS) is widely used in quantitative proteomic studies. With this technique, the relative abundance of thousands of proteins can be efficiently profiled in parallel, greatly facilitating the detection of proteins differentially expressed across samples. However, this task remains computationally challenging. Here we present a new approach, termed Model-based Analysis of Proteomic data (MAP), for this task. Unlike many existing methods, MAP does not require technical replicates to model technical and systematic errors, and instead utilizes a novel step-by-step regression analysis to directly assess the significance of observed protein abundance changes. We applied MAP to compare the proteomic profiles of undifferentiated and differentiated mouse embryonic stem cells (mESCs), and found it has superior performance compared with existing tools in detecting proteins differentially expressed during mESC differentiation. A web-based application of MAP is provided for online data processing at http://bioinfo.sibs.ac.cn/shaolab/MAP.

Protein-Driven Mechanism of Multiorgan Damage in COVID-19

“…In total 5346 human proteins were quantified in at least one sample. The authors reported the differentially expressed proteins (DEPs) determined with Model-based Analysis of Proteomic data (MAP) [ 70 ]. In this case muscle and spleen samples were not considered due to the lack of the corresponding normal tissue data in the Human Proteome Map database (see Methods).…”

Piriformospora indica and Azotobacter chroococcum Consortium Facilitates Higher Acquisition of N, P with Improved Carbon Allocation and Enhanced Plant Growth in Oryza sativa

“…Carbamidomethyl on cysteine was set as fixed modification while, for the variable modification, oxidation on Met and acetylation on protein N-terminal and phosphorylation on site (S, T, Y) were specified for database search. We set the threshold for protein false discovery rate and peptide spectrum match to 0.01. p -values for the significance were calculated by using customized python scripts following Li et al [ 40 ]. Using UniProt accession IDs as identifiers, bioinformatics analyses for rice proteins was conducted using the database for annotation, visualization, and integrated discovery (DAVID) [ 41 ].…”

Bioinformatic Analysis of Temporal and Spatial Proteome Alternations During Infections

“…DESeq (Anders and Huber, 2010) and edgeR are Bioconductor R packages suitable for differential expression analyses with relatively small proteomic datasets (Branson and Freitas, 2016). Differential expression can likewise be assessed by the generalized linear mixed-effects model (GLMM) (Choi et al, 2008), linear mixed-effects model (Hill et al, 2008), or quasi-likelihood modeling generalized linear model (GLM) (Li et al, 2019).…”

“…This occurs due to interference with quantification from co-fragmented peptides (Rauniyar and Yates, 2014). This undermines the ability of isobaric labeling to be genuinely quantitative (Li et al, 2019). Methods exist to correct for "ratio compression" (Savitski et al, 2013), but detailed discussion of the topic is outside the scope of this manuscript.…”

“…Unlike earlier algorithms that required technical replicates to determine experimental error and identify proteins with significantly altered expression, MAP uses regression analysis to calculate local error. These error estimates are then employed in the detection of proteins with significantly altered expression without the need for technical replicates to model technical and systematic errors (Li et al, 2019). In comparative analyses, MAP outperformed other, replicate-based algorithms (Zhang et al, 2010).…”