Protein fingerprinting with digital sequences of linear protein subsequence volumes: a computational study

ABSTRACT

Proteins in a proteome can be identified from a sequence of K integers equal to the digitized volumes of subsequences withL residues from the primary sequence of a stretched protein. Exhaustive computations on the proteins of Helicobacterpylori (UniProt id UP000000210) with L and K in the range 4–8 show that *90% of the proteins can be identified uniquelyin this manner. This computational result can be translated into practice with a nanopore, an emerging technology that doesnot require analyte immobilization, proteolysis or labeling. Unlike other methods, most of which focus on a specific targetprotein, nanopore-based methods enable the identification of multiple proteins from a sample in a single run. Recent workby Kennedy, Kolmogorov and associates shows that the blockade current due to a protein molecule translocating through ananopore is roughly proportional to one or more contiguous residues. The present study points to a modified version inwhich the volumes of subsequences (rather than of single residues) may be obtained by integrating the blockade current dueto L contiguous residues. The advantages arising from this include lower detector bandwidth, elimination of thehomopolymer problem and reduced noise. Because an identifier is based on near as well as distant (up to 2KL-L) residues,this approach uses more global information than an approach based on single residues and short-range correlations. Theresults of the study, which are available in a data supplement, are discussed in detail. Potential implementation issues areaddressed.