Fig. 1: Overview of FLASHIda.

a MS duty cycle control employed by FLASHIda. FLASHIda processes each MS full scan within a few milliseconds (about 20 ms on average) and optimizes the acquisition of the next cycle. b Key steps of FLASHIda. FLASHIda uses Thermo iAPI to access MS full scan in real-time. FLASHIda takes three steps to select high quality precursor isotopomer envelopes of diverse proteoforms. The first step is to transform the input m/z-intensity spectrum into a mass-quality (QScore) spectrum, and the second is to select precursors in the transformed spectrum so that the number of unique proteoform level identifications (simply proteoform ID) is maximized (Top-N QScore precursor acquisition with a quality-based exclusion list). Lastly, the charge state and isolation window size for each selected mass are dynamically determined to minimize interference from noise or coelution. The determined isolation window m/z ranges are provided to the instrument through the Thermo iAPI interface. c QScore calculation via logistic regression. After the deconvolution, the quality of the resulting fragmentation is measured for each precursor using a logistic regression with six features (see Methods). The features are extracted from the peaks in the original spectrum corresponding to the precursor. The quality metric, or QScore, is the probability estimated by the regression that the resulting fragment spectrum to be successfully identified. d Top-N QScore precursor selection with a quality-based exclusion list. With the transformed mass-QScore spectrum, FLASHIda attempts to select the high quality precursors while maximizing the identified proteoform diversity, simply by taking the top N masses with the highest QScore (Top-N QScore acquisition). To reduce the redundancy in precursor ion selection, FLASHIda conservatively excludes masses that are highly likely to be already identified; if a mass has been acquired multiple times within a short RT duration, the probability that at least one of the acquired MS2 spectra from the mass is identified is calculated with its QScores (see Methods for detail). This probability, called TQScore, is immediately updated per mass upon each acquisition, and the masses of high TQScores (>0.9) are excluded from the acquisition for a short RT duration.