Fig. 5: Algorithm selection and performance of label-free quantification.
From: AlphaPept: a modern and open framework for MS-based proteomics
a Timings of different, highly optimized solvers from the SciPy ecosystem to extract optimal protein intensity ratios in AlphaPept. Solvers showed drastic differences in speed, closeness to “ground truth”, and proportion of successful optimizations on in silico test data. Each optimization was run repeatedly (n = 200), and error bars show one standard deviation. Based on these tests, AlphaPept employs a hybrid optimization strategy that uses L-BFGS-B and Powell for optimized performance, robustness, and speed. b Uncorrected MaxQuant intensities. c Intensity distributions after MaxQuant LFQ optimization. d Comparing the AlphaPept LFQ solver on MaxQuant output data demonstrates better separation in mixed-species datasets with smaller standard deviations and more protein groups retained.
