Automatic Classification of Protein Sequences into Structure/Function Groups via Parallel Cascade Identification: A Feasibility Study

Abstract

A recent paper introduced the approach of using nonlinear system identification as a means for automatically classifying protein sequences into their structure/function families. The particular technique utilized, known as parallel cascade identification (PCI), could train classifiers on a very limited set of exemplars from the protein families to be distinguished and still achieve impressively good two-way classifications. For the nonlinear system classifiers to have numerical inputs, each amino acid in the protein was mapped into a corresponding hydrophobicity value, and the resulting hydrophobicity profile was used in place of the primary amino acid sequence. While the ensuing classification accuracy was gratifying, the use of (Rose scale) hydrophobicity values had some disadvantages. These included representing multiple amino acids by the same value, weighting some amino acids more heavily than others, and covering a narrow numerical range, resulting in a poor input for system identification. This paper introduces binary and multilevel sequence codes to represent amino acids, for use in protein classification. The new binary and multilevel sequences, which are still able to encode information such as hydrophobicity, polarity, and charge, avoid the above disadvantages and increase classification accuracy. Indeed, over a much larger test set than in the original study, parallel cascade models using numerical profiles constructed with the new codes achieved slightly higher two-way classification rates than did hidden Markov models (HMMs) using the primary amino acid sequences, and combining PCI and HMM approaches increased accuracy.

Interpreting Microarray Data and Related Applications Using Nonlinear System Identification

“…For example, consider the problem of predicting the structure/function family of a novel protein given only its primary amino acid sequence. The amino acid sequences can be regarded as the inputs and their corresponding families as the outputs [40,51]. First, some means is used to map the amino acid sequence into a corresponding numerical sequence and similarly to numerically designate the families to be distinguished.…”

“…Indeed, as reported subsequently [51], use of certain "simultaneously axially and radially aligned hydrophobicities (SARAH) scales" to uniquely encode the amino acids via 5-tuples increased PCI classification accuracy. In the SARAH1 scale, the code for each amino acid has three entries that are 0 and two that are both 1 or both 21.…”

“…For both scales, the amino acids were ranked according to the Rose scale (breaking ties), and then the codes were assigned in descending order of their binary number values. Either SARAH scale leads to either a numerical sequence five times longer than the amino acid sequence or to a set of five signals each of the same length as the amino acid sequence [51] for use with a five-input parallel cascade classifier [14]. To classify a novel protein that was encoded by a SARAH scale, an MSE ratio test [51] yielded higher accuracy than using the sign of the mean output as discussed above.…”

“…Either SARAH scale leads to either a numerical sequence five times longer than the amino acid sequence or to a set of five signals each of the same length as the amino acid sequence [51] for use with a five-input parallel cascade classifier [14]. To classify a novel protein that was encoded by a SARAH scale, an MSE ratio test [51] yielded higher accuracy than using the sign of the mean output as discussed above. In the ratio test, the MSE of the output signal from the ideal value for a class is normalized by the MSE of the training exemplars for that class, and the sequence is assigned to the class with the smallest of the ratios.…”

“…In the ratio test, the MSE of the output signal from the ideal value for a class is normalized by the MSE of the training exemplars for that class, and the sequence is assigned to the class with the smallest of the ratios. When tested on over 16,000 protein sequences in two-way classifications, parallel cascade classifiers using SARAH1-encoded sequences outperformed state-of-the-art hidden Markov models [53] trained with the same three protein exemplars (one sequence each from globin, calcium-binding, and kinase classes) [51].…”