Databases profile methods

Because domains can be considered independent structural and functional units, each domain can be analyzed independently once it has been determined that the query protein contains more than one domain. The identification of functional domains can be performed directly by matching the entire query sequence or a portion of it to a profile from a domain database. Alternatively, the existence of functional domains can be evaluated through indirect inference. For instance, if the query protein contains a well-characterized domain that matches a database profile and the rest of the sequence is not covered by any known domain, that uncovered region (provided it has a reasonable length) can be assumed to contain an additional domain. For cases in which there are no matches to domains or protein families in databases, the existence of multiple domains in the protein of interest can still be inferred through other methods. For example, the connectors between domains tend to be disordered or flexible linkers. Accordingly, predictions of disorder or composition bias, linker predictions, or secondary-structure predictions can be used to infer the spatial location of uncharacterized domains. 

Since paralogs are typically more distantly related than orthologs, their detection frequently requires sensitive protein comparison methods such as profiles or HMMs. Even the detection of orthologs can, under some circumstances, require sophisticated database searching methods, e.g. if the corresponding organisms belong to very distant phyla. 

In this article we have introduced the Bruker SGF Profiling method for the authentication, verification and quality control of fiuit juices. In addition to tihe quantification of a large array of characteristic compounds, this fully automated NMR screening technique uses statistical models for the estimation of fhiit content or the origin of the juice. This analysis tool can show known and unknown deviations from normality. Currently, routines are under development to identify unknown deviations by constructing spectral patterns which can be compared to an existing reference compound database. ... 

Figure 18.1 Schematic representation for the prediction of the complete metabolic profile of a molecule using databases and a machine learning approach. In this example various metabolite rules are used to illustrate how this method will be implemented. Molecule B could also represent metabolites derived from Molecule A.

The DFT/COSMO calculations are the rate-limiting part of the method and can easily take a few hours for molecules with up to 40 heavy atoms on a 3-GHz computer [36]. To overcome speed limitations, the authors developed the COSMOfrag method. The basic idea of this method is to skip the resource-demanding quantum chemical calculations and to compose a profiles of a new molecule from stored a profiles of precalculated molecules within a database of more than 40000 compounds. A comparison of the full and fragment-based versions for log P prediction was performed using 2570 molecules from the PHYSPROP [37]. RMSE values of 0.62 and 0.59 were calculated for the full COSMO and COSMOfrag methods, respectively [36]. 

Within the project we also evaluated alternative methods as tools to obtain information on the toxicological and physicochemical profile of the pollutants. In this paragraph, an example of the application of QSARs models is reported a comparison is done between predicted values from different models or between QSARs evaluation and experimental values from internationally recognized databases. 

These in vivo and in vitro human metabolism studies indicate that pyrethroids undergo rapid metabolism and elimination as observed in rats, and qualitative metabolic profiles (e.g., kinds of metabolites) of pyrethroids are assumed to be almost the same between humans and rats, suggesting that a large database of animal metabolism of pyrethroids could provide useful information for the evaluation of behavior of pyrethroids in humans. Nowadays, human pesticide dosing studies for regulatory propose are severely restricted in the US, and thus detailed comparison of in vitro metabolism (e.g., metabolic rate constants of pathways on a step-by-step basis) using human and animal tissues could be an appropriate method to confirm the similarity or differences in metabolism between humans and animals. 

Prosite is perhaps the best known of the domain databases (Hofmann et al., 1999). The Prosite database is a good source of high quality annotation for protein domain families. Prosite documentation includes a section on the functional meaning of a match to the entry and a list of example members of the family. Prosite documentation also includes literature references and cross links to other databases such as the PDB collection of protein structures (Bernstein et al., 1977). For each Prosite document, there is a Prosite pattern, profile, or both to detect the domain family. The profiles are the most sensitive detection method in Prosite. The Prosite profiles provide Zscores for matches allowing statistical evaluation of the match to a new protein. Profiles are now available for many of the common protein domains. Prosite profiles use the generalized profile software (Bucher et al., 1996). 

The SBASE database is a collection of annotated protein sequence segments (Murvai et al., 1999). SBASE avoids using consensus methods such as profile-HMMs and uses pairwise methods to detect domains. The database includes more than 130,000 annotated sequence segments that have been clustered into groups on the basis of BLAST similarities. SBASE currently contains 1038 domain families. 

---