Abstract 207: Reliable identification of mutations in bottom-up proteomics

Abstract

Abstract Knowledge of missense mutations is of significant biological importance as it provides valuable insights into alterations of phenotype. While identification of reference proteins is routinely performed in proteomics, identification of mutations remains still fairly uncommon. Well-established methods for reference peptides are generally insufficient for altered ones; e.g., utilizing mutant-augmented database search usually results in inadmissible proportion of false positives. This undesirable situation is, however, rather natural consequence of peptide fragmentation, its computational modelling and evaluation of their correspondence. Often, there are many peptides having the same or similar agreement with acquired spectrum, thereby preventing agreement-based decision. These situations happen frequently in identification of mutant peptides (e.g., homologous peptides with PTMs, semi-specific peptides), therefore their reliable identification requires additional treatment. To deeper understand properties of identification, we have formally studied generalized version of the identification problem and derived its optimal solution under particular assumptions. In proteomics, the selection of optimal solution can be always found in finite time; moreover, the strategy is straightforward to implement, enables parallelization and provides guarantees over claimed interpretations. The solution utilizes almost exclusively spectral data of product ions without additional LC/MS information. In practice, however, it is beneficial to employ precursor isotopic distribution analysis for correction of non-monoisotopic selection of precursor as that would otherwise systematically result in artefacts. The behaviour of proposed method was validated in a variety of scenarios that could be broadly categorized as direct (known spectral content) and indirect (partial knowledge of sample content). For direct validation, synthetic combinatorial peptide library of 400 peptides (all coded amino acids in 12. and 13. codon of KRAS peptide) was measured and interpreted; the sensitivity and specificity obtained was 0.81 and 0.99 respectively. Indirect validation was, at first, performed on in-house samples with RNA-Seq data available (HCT116), giving 97 missense RNA-Seq supported mutations. Similarly, the analysis was done on proteome data of NCI60 cell lines, resulting on average in 23 SNVs with maximum of 56 SNVs (HCC-2998). Identified alterations correspond, in general, to highly expressed genes. In summary, the approach was shown to reliably identify missense mutations in a range of cancer cell lines; its wide applicability enhances interpretation of standard bottom-up proteomics data. Citation Format: Miroslav Hruska, Lakshman Varanasi, Jiri Voller, Petr Dzubak, Marian Hajduch. Reliable identification of mutations in bottom-up proteomics [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 207. doi:10.1158/1538-7445.AM2017-207