Amino acid sequence assignment from single molecule peptide sequencing data using a two-stage classifier

Abstract

We present a machine learning-based interpretive framework (whatprot) for analyzing single molecule protein sequencing data produced by fluorosequencing, a recently developed proteomics technology that determines sparse amino acid sequences for many individual peptide molecules in a highly parallelized fashion [1] [2]. Whatprot uses Hidden Markov Models (HMMs) to represent the states of each peptide undergoing the various chemical processes during fluorosequencing, and applies these in a Bayesian classifier, in combination with pre-filtering by a k-Nearest Neighbors (kNN) classifier trained on large volumes of simulated fluorosequencing data. We have found that by combining the HMM based Bayesian classifier with the kNN pre-filter, we are able to retain the benefits of both, achieving both tractable runtimes and acceptable precision and recall for identifying peptides and their parent proteins from complex mixtures, outperforming the capabilities of either classifier on its own. Whatprot's hybrid kNN-HMM approach enables the efficient interpretation of fluorosequencing data using a full proteome reference database and should now also enable improved sequencing error rate estimates.

Beam search decoder for enhancing sequence decoding speed in single-molecule peptide sequencing data

“…The values of the parameters are the same as the ones used in [14]. This paper assumes the same parameters along all colors, as was also done in [14], but the method can be easily extended to color-specific parameters. [14].…”

“…This paper assumes the same parameters along all colors, as was also done in [14], but the method can be easily extended to color-specific parameters. [14].…”

“…Every amino acid can be represented with a letter; then, the peptide can be represented by a sequence of letters. [14] defines the dye sequence representation of a peptide by representing each amino acid by the index of the color that can attach to it. If no dye attaches to that amino acid, a "."…”

“…The proposed beam search decoder has a user-set variable N B which is the number of most likely states kept at each cycle. Our decoder achieves much faster computation times and also obtains slightly higher accuracy than the hybrid model of [14]. Although probeam has a poorer estimation of the probability of the picked sequence, it can achieve 100x faster computing times when decoding using a dataset of twenty thousand proteins.…”

“…Smith et al introduce in [14] a model-based decoder to obtain the most likely peptide given the light intensity reads. First, Smith et al propose a HMM that considers all the significant failures during the process and has optimal accuracy.…”