Data-independent acquisition of peptide mass spectrometry data has the potential to enable improved quantitative reproducibility for plasma biomarker studies. One approach for peptide identification is to take advantage of existing peptide spectral libraries, as these can be merged with a local seed library to make an extended reference library using software such as SWATHXtend, as we have previously reported. Important to recognise when merging libraries is that the concomitant larger extended library yields increased probability of false-positive extraction. In this study, we explored optimising plasma SWATH library generation aiming to maximum protein coverage, while minimising false-positive detections.
We used a locally acquired plasma library as a seed to make two extended libraries by merging spectral data downloaded from the plasma dataset published by Liu et al  (1885 proteins) and from the human SWATH library in SWATHAtlas after selecting for plasma proteins reported in the HPP-2017 update  (3286 proteins). Data was acquired on a TripleTOF 6600 using 60min LC SWATH runs from five human plasma samples. We used PeakView for peptide extraction with protein FDR set at 99% confidence. Combining only the proteins detected from the two extended libraries with our local seed library, we obtained a new SWATH plasma library containing 1161 proteins. This library is 38% and 65% smaller than the original extended two libraries respectively. It is more specific to the plasma samples therefore detected more proteins with fewer false positives than using any of the individual local or extended reference libraries.