ENSEMBL database

The main characteristics of Prous Ensemble database are listed as follows ... 

All structures were selected from the Ensemble database. In addition to approved therapeutic agents these particularly include a set of lead compounds entered in advanced clinical/preclinical trials (a total of 16,540 compounds). Structures were extracted according to the assigned activity class, where the class indicates a common target-specific group such as GPCRs, kinases and proteases, nuclear receptors, and ion channels as well as more than 150 subclasses (for example, serotonin, tachykinin and dopamine receptors, tyrosine, Abl, Aurora and serine/threonine kinases, cysteine and serine proteases, etc.). Prior to the statistical experiments, the molecular structures should be filtered and normalized in order to fulfill certain criteria (see Subheading 3.5 and 4). 

In the first step, the selected dataset including 16,540 structures (see Subheading 2.2) was exported from the Ensemble database into the corresponding SDF file, then it was imported into the SmartMining software thus, the internal database was fully formed except 36 structures with unrecognized structures. The import procedure was completed in less than 2 min. 

Fig. 1. CONREAL sequence input form. (A) Two sequences in Fasta format can be pasted into the text field or (B) provided in a plain text file or (C) sequences can be automatically retrieved from the Ensembl database using a gene name or keyword and a species name.

Coordinates provided in the results section are absolute coordinates with the +1 position corresponding to the beginning of the submitted sequences. Note that these coordinates are different from coordinates used to retrieve sequences from Ensembl database at Subheading 3.1., step 5. 

Using the ENSEMBL database, it is possible to classify regions of the genome as related to genes. This information was used to investigate the number of... 

ISO EN 9920 provides a large database of thermal insulation values, which have been measured on a standing thermal manikin. One set of tables gives the insulation values for a large number of ensembles (Table 6.14).. An-f)ther set of tables gives insulation values for individual garments (Table 6.15), on the basis of which the insulation for a whole ensemble can be estimated. 

It is important to emphasize that this lattice database is highly idealized compared to real databases. Unlike the lattice database, real databases cannot be treated as thermodynamic ensembles of protein-ligand complexes equilibrated at room temperature [33,34]. Two of the more straightforward reasons are mentioned here. First, real databases are inherently biased toward strong binders (K < 10 pM), because weak binders are difficult to crystallize and of lesser interest. Second, as mentioned above, real databases are not composed of a representative selection of proteins and ligands, and their compositions are biased toward peptide and peptidomimetic inhibitors and certain protein superfamilies. In contrast, because only one protein and four ligand types are used, the lattice database should have representative ligand compositions. 

In the case of being successful in calculating multiple conformations by using time- or ensemble-averaged MD restraints the solved molecular structures are presented as 3D models and can be deposited in an electronic structure database (17). Finally, it is recommended to provide an accurate explanation of the procedures used for the structure elucidation because the application of different methods (NMR, DG, MD, SA, Monte-Carlo calculations. X-ray crystallography) may result in varying conformational models which do not implicitly display the real state of a molecule. This aspect should be always kept in mind when dealing with structure determination methods. 

Ensemble, [Internet], URL http //www.prous.com/ databases/ensemble/ensemble.html, accessed 7-30-2000. 

The ensemble of ESTs in the available databases should cover all genes of the genome and all parts of each gene. At present, there are about 1.3 million human ESTs covering about 75,000 human mRNAs. Thus one mRNA is hit on the average by >10 ESTs, but one EST can cover only a fraction of mRNA sites (about 300 nt per about 2000 sites). 

For fuzzy virtual screening purposes, highly active molecules with different scaffolds are combined into an MTree model. By combining the information of remotely related actives into a single model, efficient database searches with molecule ensembles are possible. 

---