Fft based template matching

This r. Considering that the models have been built from fragment libraries designed to match fragments from known proteins within about 0.

It seems possible that even closer agreement might be achieved by using larger fragment libraries, but this would come at the expense of more time spent examining the fits of fragments to the map.

Alternatively, the agreement could be improved by refinement of the models that are obtained. The author is grateful to the NIH for generous support. National Center for Biotechnology Information , U. Acta Crystallogr D Biol Crystallogr. Published online Dec Thomas C. Author information Article notes Copyright and License information Disclaimer.

Correspondence e-mail: vog. Received Jul 1; Accepted Oct 1. This article has been cited by other articles in PMC. Abstract An algorithm for the automated macromolecular model building of polypeptide backbones is described.

Keywords: model building, template matching, fragment extension. Introduction Model building is a key and often time-consuming step in macromolecular structure determination.

Fragment libraries Four fragment libraries were constructed. Segment extension Construction of a segment of a polypeptide chain is accomplished by iterative fragment extension.

Chain assembly The procedures described above generate a set of segments that may correspond to portions of polypeptide chain. Results and discussion 3. Open in a separate window. Figure 1. Figure 2. Figure 3. Tests with structures solved by MAD and SAD The procedure for automated main-chain model-building described here was further tested by applying it to a set of eight experimental maps with varying resolution, quality figure of merit and number of residues in the asymmetric unit.

Acknowledgments The author is grateful to the NIH for generous support. References Berman, H. Nucleic Acids Res.

Cowtan, K. Acta Cryst. D 54 , — D 57 , — A 33 , 13— Natl Acad. USA , 99 , — Methods Enzymol. Protein Sci. D 56 , — Cell , 90 , — EMBO J. A 47 , — D 53 , — D 49 , — Morris, R. D 58 , — Structure , 6 , — Nature Biotechnol. Nature Struct. Add a comment. Active Oldest Votes. You are using the Fourier Transform to calculate the cross correlation; it's as simple as that.

Consider using phase correlation, the extra effort is very small and you can get great results. Improve this answer.

I had to edit my answer; image padding can be used to interpolate correlation results, which is another problem. Generally, when matching a smaller template to an image then the template is padded and the image does not need to be padded. Matthias Odisio Matthias Odisio 2, 12 12 silver badges 19 19 bronze badges. Sign up or log in Sign up using Google. Sign up using Facebook. Sign up using Email and Password.

Post as a guest Name. Email Required, but never shown. The Overflow Blog. Podcast Helping communities build their own LTE networks. Podcast Making Agile work for data science. In its simplest form, this implies solving the following optimization:. A common choice of similarity metric is cosine similarity , which is defined as:. Cosine similarity is a reliable metric, but as defined above its evaluation is rather costly. Depending on image and template sizes, the intended use case e.

Cosine similarity template matching can be sped up with application of the convolution theorem. It uses the popular OpenCV library, with some further optimizations particular to its implementation of the Fourier transform, explained in the comments.