Skip to main content
. 2020 May 7;3:218. doi: 10.1038/s42003-020-0938-9

Fig. 7. Evaluation of selective removal of variation due to batches.

Fig. 7

Batch registration of the SDY420 study was performed by registering each of the internal standards to a reference (Batch 4) internal standard to generate 24 batch-specific registration templates, and the shifts in these templates were applied to the appropriate batches of individual samples (total 260 samples). For three batches, consensus reference samples were used because internal standards were unavailable. A concatenate of the registered samples was clustered in SWIFT, and all samples were assigned to the resulting cluster template. QC plots (described in Fig. 1) show the variation in the centroids of 13 selected channels of each cluster in all internal standards before and after registration (a) and in 12 actual samples from one batch before and after batch registration (b). Arrows indicate samples that are outliers for CD20 or CD45RA cluster centroids.