Extremely Fast and Accurate Open Modification Spectral Library Searching of High-Resolution Mass Spectra Using Feature Hashing and Graphics Processing Units

Abstract

Open modification searching (OMS) is a powerful search strategy to identify peptides with any type of modification. OMS works by using a very wide precursor mass window to allow modified spectra to match against their unmodified variants, after which the modification types can be inferred from the corresponding precursor mass differences. A disadvantage of this strategy, however, is the large computational cost, because each query spectrum has to be compared against a multitude of candidate peptides. We have previously introduced the ANN-SoLo tool for fast and accurate open spectral library searching. ANN-SoLo uses approximate nearest neighbor indexing to speed up OMS by selecting only a limited number of the most relevant library spectra to compare to an unknown query spectrum. Here we demonstrate how this candidate selection procedure can be further optimized using graphics processing units. Additionally, we introduce a feature hashing scheme to convert high-resolution spectra to low-dimensional vectors. Based on these algorithmic advances, along with low-level code optimizations, the new version of ANN-SoLo is up to an order of magnitude faster than its initial version. This makes it possible to efficiently perform open searches on a large scale to gain a deeper understanding about the protein modification landscape. We demonstrate the computational efficiency and identification performance of ANN-SoLo based on a large data set of the draft human proteome. ANN-SoLo is implemented in Python and C++. It is freely available under the Apache 2.0 license at https://github.com/bittremieux/ANN-SoLo .

Large-scale tandem mass spectrum clustering using fast nearest neighbor searching

“…Instead, a feature hashing scheme [45] is used to convert high-resolution MS/MS spectra to lowdimensional vectors while closely capturing their fine-grained mass resolution. The following two-step procedure is used to convert a high-resolution MS/MS spectrum to a vector (figure 1A) [4]: Convert the spectrum to a sparse vector using small mass bins to tightly capture fragment masses. Hash the sparse, high-dimensional, vector to a lower-dimensional vector by using a hash function to map the mass bins separately to a small number of hash bins. …”

“…Instead, a feature hashing scheme [44] is used to convert high-resolution MS/MS spectra to lowdimensional vectors while closely capturing their fine-grained mass resolution. The following twostep procedure is used to convert a high-resolution MS/MS spectrum to a vector (figure 1A) [4]:…”

“…By making use of advanced algorithmic techniques, falcon is optimized for highly efficient spectrum clustering. It uses feature hashing to convert high-resolution MS/MS spectra to low-dimensional vectors [4], in combination with efficient nearest neighbor searching in the vector metric space using the cosine similarity [5]. Next, spectra are grouped into clusters by density-based clustering [8].…”

Emerging mass spectrometry-based proteomics methodologies for novel biomedical applications

“…Similarly, methods utilizing prior knowledge of the glycan structures released from glycoproteins are supporting glycomics studies [65]. A new informatic approaches called 'open modification searching' (OMS) focus not on specific predefined PTMs, but rather it detects shifts in peptide and fragment ion masses, allowing for the discovery of any possible PTM present in the data [66][67][68]. Although a powerful approach, OMS analyses often remain difficult to interpret biologically.…”

Discovery of protein modifications using high resolution differential mass spectrometry proteomics

“…Finally, recent implementations of fragment ion indexing and other efficiency concepts support fast assignment of modified mass spectra (Kong et al 2017;Devabhaktuni et al 2019;Bittremieux et al 2019;Chi et al 2018;Kahnert et al 2020).…”