TAMPOR Initial Release

Well tested with batched protein quantification data (untransformed abundance), from either tandem mass tag (TMT) or label free quantification (LFQ) bottom-up proteomics. Also tested with FPKM or TPM transcriptomic data. Both build biologically meaningful coexpression networks (e.g. WGCNA) which were latent and undetectable without robust batch effect removal by TAMPOR.

Preprocessing of FPKM data is required to handle large numbers of zero-read measurements. Zeroes can be replaced with NA before input to TAMPOR. Assumptions of this preprocessing are that no batch used for analysis has both high transcript measurements AND a preponderance (>50%) zero values. Rows with greater than or equal to 50% missing values are thrown out during cleanup.

Data sets were rescued when R sva package ComBat function was insufficient in removing enough batch artifact to reproduce coexpression networks, but this observation needs further testing and validation.