Decoding the Impact of Isolation Window Selection and QuantUMS Filtering in DIA-NN for DIA Quantification of Peptides and Proteins

Proteomic studies using data-independent acquisition
(DIA) have
gained momentum in all fields of biology. Search engines are evolving
to keep up with the latest developments in instrument technology.
DIA-NN is the most popular software for DIA analysis under an academic
use license. The QuantUMS algorithm in DIA-NN improves quantification
quality control by calculating three scores (protein group MaxLFQ
quality, empirical quality, and quantity quality) that assess the
agreement between MS1 and MS2 features. Here, we show that applying
specific cutoffs to these scores can significantly impact the results.
To enable you to make a more informed decision about what represents
a reasonable trade-off (identification and quantification), we evaluated
the impact of different combinations of the scores on data acquired
using different isolation windows and a mixture of two species with
a known ratio. To test consistency and reproducibility across the
six different versions of DIA-NN, we compared them and found high
reproducibility except for version 1.9. We show that filtering by
QuantUMS scores removes proteins with low abundances and high coefficients
of variation. Finally, we developed the QC4DIANN Shiny application
in the R language for interactive quality control automation.