Batch effect refers to technical variation or non-biological differences between measurements of different groups of samples. In metaproteomics, a performance change of LC-MS/MS could lead to batch effects.
We wrapped this app to help diminish the batch effects between different batch of runs. Please not that different batches should be distributed as a balanced experimental design in order to get a valid output. It is crucial that you randomize your sample before sample analysis.
About the app
This app is a wrapper of the exploBATCH, an R package for evaluating and correcting for batch effect in genomic studies. Please read the following refererence for more details on exploBATCH (Sci Rep. 2017 Sep 7;7(1):10849. doi: 10.1038/s41598-017-11110-6. A Novel Statistical Method to Diagnose, Quantify and Correct Batch Effects in Genomic Studies.) , and
refer to https://github.com/syspremed/exploBATCH for more details about the package. The comBat method is decribed in SVA package. Here we used the parameter option.
This app requires the input of two tables: data matrix with sample names and intensities of each feature, and meta file specifying which batch is a sample belong to.
Remove batch effect
To begin, select the analysis type. Please note that correctBatch is slow for large datasets.
Then, specify parameters for the correction and click “Go analysis”.
A box will show up to let you know the progress. Then, PCA or PPCCA will display to allow you compare sample distribution before and after correction.
By switching to “More supporting plots”, you can also check the proportion of PCA variations across all PCs from PCA analysis. There is for sure some variation between the two method correction, but usually highly consistent to each other.
Finally, choose result table to view/download for your further analysis.