Fig. 1: Overview of the alphaDIA framework.
From: AlphaDIA enables DIA transfer learning for feature-free proteomics
a, Components of alphaDIA and the integration into the alphaPept ecosystem. AlphaDIA uses alphaRaw and alphaTims50 for accessing raw data from all major vendors. Importing and prediction of spectral libraries are facilitated by alphaBase and alphaPeptDeep20. After successful search, LFQ is performed using directLFQ40. Two leftmost mass spectrometry instrument illustrations created with BioRender. b–f, TIMS DIA data acquired using optimal dia-PASEF39 are searched using a peptide-centric algorithm. b, The library entry for a single peptide sequence is selected for search. c, Fragment spectra containing the precursor of interest are extracted and converted into a dense matrix in spectrum space. d, Information from fragments mapping to the precursor of interest are combined in a continuous score. ML, machine learning. e, AlphaDIA defines candidate peak groups with discrete integration boundaries (top row: intensities, bottom row: mass deviation from theoretical mass). f, Aggregating signal across the integration boundaries in ion mobility and retention time reveals the peptide spectrum. For further scoring, AlphaPeptDeep spectrum predictions are used.
