Similarity searching local measures

Local Measures for Distance-Based 3D Similarity Searching... 

Asymmetric simhaiity measures allow fuzzy super- and substructure searching. A substructure search is defined as looking for structures containing the given query and a superstructure search is defined as looking for structures embedded in the given query. In both cases asymmetric local similarity is estimated. 

The most popular tool currently in use is BLAST (Basic Local Alignment Search Tool) (3 7) from the NCBI. BLAST is an example of a heuristic that attempts to optimize a specific similarity measure. The most recent revisions to the algorithm are gapped BLAST and PSI-BLAST (38), with improved accuracy for PSI-BLAST using composition-based statistics... 

When searching for local similarities of two molecules, the decoded local shape matrices Ib(a,b,Mi) of molecule M) are compared to various diagonal blocks of the global shape matrix s(a,b,M2) of molecule M2. In the most general case, the local shape matrix Ib(a,b,Mi) is used as a template, and it is compared to k-dimensional blocks of s(a,b,M2) obtained by all possible simultaneous row and column permutations. If the size ordering is considered important then only those permutations are taken which preserve the monotonicity of size ordering in the permuted diagonal block that is compared to the template. A local similarity measure... 

The present authors have implemented a software tool called oplAnalyzer in MATLAB for MALDI-MS data preprocessing. To detect peaks in a mass spectrum, a search is made for locations of maximal value within a local m/z window, the size of which is 11 discrete sampUng points. This method is similar to another peak detection method [35]. For alignment, peak detection is applied to the mean spectra, analogous to the method of Karpievitch and others [34]. Subsequently, the peaks in an individual spectrum are aligned to this set of common peaks. For each common peak, its value in an individual spectrum is that of the closest detected peak in that spectrum if the distance between the common peak and the closest peak (in the m/z axis) is less than Vs Da. (A better choice is based on the actual mass accuracy of the measurement) If there is no such peak, then the m/z of the... 

Fingerprints a metric of structural similarity In a global structure search, very similar structures always appear frequently. Duplicate structures do not only create inconvenience in postprocessing, but also lead to the situation that the search is trapped in some local minimum but not the ground state. Therefore, a technique to measure the similarities between structures is needed. In USPEX, we use the so-called fingerprint function to describe a crystal structure. It is very similar to pair distribution function (PDF), which for an elemental solid is ... 

---