Prediction targets of CASP4

Throughout this Critical Assessment of Protein Structure Prediction (CASP) special issue of Proteins, the CASP target numbers T0086–T0128 are used, as well as occasional references to numbers from previous CASP experiments. This article summarizes useful information about the targets in one place. Figure 1 shows thumbprint images of the targets. Table I gives details of the domain structure, name, species, and Protein Data Bank (PDB) identifier (if available), of the target. It also provides a brief description of the structure within the context of our existing knowledge and classification of protein structures, such as in the Structural Classification of Proteins (SCOP) database. Where relevant, possible evolutionary relationships are given, with a notation as to whether related structures were available to predictors or have only become available since CASP4. In the case of comparative modeling targets, a description is given only where there are notable structural differences; the reader is referred to the more detailed description of these targets in the article of the assessor. Targets in CASP are divided roughly into three categories, comparative modeling (CM); fold recognition (FR), and new folds (NF), as described in the Introduction to the CASP experiment. The boundaries between these categories are naturally blurred. To address this issue in CASP4, targets have been classified into six categories instead of three, the extra categories reflecting this overlap between the original three categories. Categories were initially drawn up by the fold recognition assessor (in whose article the issue is discussed) and refined in discussion with the other assessors. The classification was made at the domain level. Of the 40 targets, 13 are multiple domain; of these, 8 have domains in different categories.