Profile searching in nucleic acid sequences using the fast fourier transformation

摘要

One embodiment of the present invention provides methods for detecting known blocks of functionally aligned protein sequences in a test nucleic acid sequence, e.g., in an uncharacterized EST. The method can include the following steps. A) Reverse translate the set of protein sequences to a set of functionally aligned nucleic acid sequences using codon-usage tables and create a profile from the set of functionally aligned nucleic acid sequences. B) Construct a first indicator function for the profile. The first indicator function corresponds to adenine. The first indicator function allows the value at a given position to be continuous between 0 and 1 as a function of the percentage presence of adenine at a particular position. C) Construct a second indicator function for the test nucleic acid sequence. The second indicator function also corresponds to adenine. D) Compute the Fourier transform of each of the indicator functions. E) Complex conjugate the Fourier transform of the second indicator function. F) Multiply the Fourier transform of the first indicator function and the complex conjugated Fourier transform of the second indicator function to obtain a Fourier transform of the number of matches of adenine bases. G) Repeat steps B-F above for guanine, thymine, and cytosine. H) Sum the Fourier transforms of the number of matches for each base, respectively, to obtain the total Fourier transform. I) Compute the inverse Fourier transform of the total Fourier transform to obtain a complex series. J) Take the real part of the series to determine the total number of base matches for the variety of possible lags of the profile relative to the test sequence. The method can then detect the presence of known blocks of functionally aligned protein sequences in a test nucleic acid sequence based on the total number of base matches for the variety of possible lags.