Quantitative Proteome-Property Relationships (QPPRs). Part 1: finding biomarkers of organic drugs with mean Markov connectivity indices of spiral networks of blood mass spectra.

Numerical parameters of the molecular networks, also referred as Topological Indices or Connectivity Indices (CIs), have been used in Bioorganic and Medicinal Chemistry to find Quantitative Structure-Activity, Property or Toxicity Relationship (QSAR, QSPR and QSTR) models. QSPR models generally use CIs as inputs to predict the biological activity of compounds. However, the literature does not evidence a great effort to find QSAR-like models for other biologically and chemically relevant systems. For instance, blood proteome constitutes a protein-rich information reservoir, since the serum proteome Mass Spectra (MS) represents a potential information source for the early detection of Biomarkers for diseases and/or drug-induced toxicities. The concept of mass spectrum network (MS network) for a single protein is already well-known. However, there are no reported results on the use of CIs for a MS network of a whole proteome to explore MS patterns. In this work, we introduced for the first time a novel network representation and the CIs for the MS of blood proteome samples. The new network bases on Randic's Spiral network have been previously introduced for protein sequences. The new MS CIs, called here Spiral Markov Connectivity (SMC(k)) of the MS Spiral graph can be calculated with the software MARCH-INSIDE, combining network and Markov model theory. The SMC(k) values could be used to seek QSAR-like models, called in this work Quantitative Proteome-Property Relationships (QPPRs). We calculate the SMC(k) values for 62 blood samples and fit a QPPR model by discriminating proteome MS, typical of individuals susceptible to suffer drug-induced cardiotoxicity from control samples. The accuracy, sensitivity, and specificity values of the QPPR model were between 73.08% and 87.5% in training and validation series. This work points to QPPR models as a powerful tool for MS detection of biomarkers in proteomics.