Multienzyme deep learning models improve peptide de novo sequencing by mass spectrometry proteomics

Abstract

Generating and analyzing overlapping peptides through multienzymatic digestion is an efficient procedure for de novo protein using from bottom-up mass spectrometry (MS). Despite improved instrumentation and software, de novo MS data analysis remains challenging. In recent years, deep learning models have represented a performance breakthrough. Incorporating that technology into de novo protein sequencing workflows require machine-learning models capable of handling highly diverse MS data. In this study, we analyzed the requirements for assembling such generalizable deep learning models by systemcally varying the composition and size of the training set. We assessed the generated models’ performances using two test sets composed of peptides originating from the multienzyme digestion of samples from various species. The peptide recall values on the test sets showed that the deep learning models generated from a collection of highly N- and C-termini diverse peptides generalized 76% more over the termini-restricted ones. Moreover, expanding the training set’s size by adding peptides from the multienzymatic digestion with five proteases of several species samples led to a 2–3 fold generalizability gain. Furthermore, we tested the applicability of these multienzyme deep learning (MEM) models by fully de novo sequencing the heavy and light monomeric chains of five commercial antibodies (mAbs). MEMs extracted over 10000 matching and overlapped peptides across six different proteases mAb samples, achieving a 100% sequence coverage for 8 of the ten polypeptide chains. We foretell that the MEMs’ proven improvements to de novo analysis will positively impact several applications, such as analyzing samples of high complexity, unknown nature, or the peptidomics field.

Multi-Modal Mass Spectrometry Identifies a Conserved Protective Epitope inS. pyogenesStreptolysin O

“…Peptide candidates were generated using three deep-learning-based de novo peptide tools: PointNovo (51), CasaNovo (52), and InstaNovo (53). For PointNovo, two in-house multienzyme-trained models (54) were utilized, whereas default models were employed for CasaNovo and InstaNovo. The study considered twelve fragment ions: a+1, a+2, b+1, b+2, y+1, y+2, a-H2O, b-H2O, y-H2O, a-NH3, b-NH3, and y-NH3, with candidate selection based on a 20 ppm tolerance at both MS1 and MS2 levels and the observation of a minimum of four fragment ions.…”

“…The highest-ranked peptide candidates were those that not only showed maximal overlap but also robust MS evidence supporting the target protein sequences. The positional confidence score (54) for the assembled heterodimeric chains of the Fab domain is illustrated in Fig. 1F .…”

Accounting for digestion enzyme bias in Casanovo

“…Moreover, this approach does not allow the model to generalize to settings where an MS/MS experiment makes use of novel combinations of digestion enzymes. Gueto-Tettay et al experiment with both of these approaches, first training models on data generated using a single enzyme and then training on multi-enzyme datasets to increase their models' generalizability [13].…”

“…In the example above, “PEPTIDKE” becomes the preferred sequence under digestion by gluC instead of trypsin. Therefore, the accuracy of these data-driven de novo sequencing models suffers when they are applied to non-tryptic data [12, 13].…”

“…Gueto-Tettay et al . experiment with both of these approaches, first training models on data generated using a single enzyme and then training on multi-enzyme datasets to increase their models’ generalizability [13].…”