An excerpt from a PDB file.

Flowchart for homology modeling with MolIDE. Step numbers to the right of each step correspond to the protocol described in the text, beginning with Step 5.

Image of model of Bateman domain of human cystathionine beta synthase with S-adenosyl methionine (SAM). Because the template, 1PVM, does not contain SAM or any other nucleotide, the model of CBS was superimposed on PDB entry 2YZQ (Kanagawa et al., unpublished), which does contain SAM. The side chains of the model were then remodeled as described in Step 14. Side chains near SAM are shown in stick representation. SAM is colored magenta. The backbone ribbon is colored from blue to red from N to C terminus. The image was produced with PyMOL (W. L. Delano, www.pymol.org).

Manual editing of a target-template sequence alignment. The unedited alignment is shown at left, and the partially edited alignment is shown at right. In each image, the alignment shows the target sequence and its predicted secondary structure above the template sequence and its experimental structure, with the same color coding as in Figure 2. The original alignment (left) shows a gap in the middle of an alpha helix. To edit the alignment, a ctrl-click on the gap in the left figure removes the gap. Shift-click on a position six residues to the right inserts a gap, and produces the alignment at right. The editing does not reduce the similarity of the aligned residues, and is more consistent with structural changes in homologous proteins.

Viewing and sorting the list of templates. The hits shown are from a PSI-BLAST search of pdbaa from profiles built from searching NCBI's nr database. The query is the same as in Figure 2, human CBS. The top window is sorted by hit number (the order in the PSI-BLAST output), and shows that there is an experimental structure of human CBS that covers residues 1 to 413 of the query (length 551). PDB entry 1JBQ is a longer structure than PDB entry 1M54. The bottom window is sorted by starting residue in the query (A=ascending order; D=descending order). The window shows that there are a number of templates for the C-terminal regulatory Bateman domain. From this window, target-template alignments can be viewed by double-clicking on the hit number in the first column (“No.”). Some Bateman domains in the PDB are insertions into inosine monophosphate dehydrogenase (IMPDH). The Bateman domains in these structures are largely disordered. This is evident once the target-template alignment is opened up. After looking through this list, and examining them visually, the PDB entry 1PVM51 appears to be the best template for the Bateman domain of CBS. It has only 2 gaps in the alignment, and is a high-resolution structure (1.9 Å) with no missing residues in the aligned region.

Viewing PSI-BLAST output from the non-redundant sequence database search and secondary structure predictions with PSIPRED within MolIDE. The target sequence file remains open (upper left) and the table of PSI-BLAST results from the non-redundant database is shown (upper right). This table can be sorted by clicking on any of the headers at the top of the table, once for ascending order (A) and once again for descending order (D). The secondary structure predictions are shown in the lower window. Predictions are shown in red (helix), green (sheet strand), and coil (gray). The intensity of the color is in proportion to the confidence values (from 0 to 9) given by PSIPRED. The region shown is for the C-terminal Bateman domain49 of the protein cystathionine beta synthase (CBS). Each line contains a secondary structure prediction from a round of PSI-BLAST. While the longer helices are well predicted, the beta sheet strands change as more remote homologues are added to the multiple sequence alignment used by PSI-BLAST (top to bottom). The beta sheet strand around 510 is predicted in the first 3 rounds but fades out at round 4. In fact, the early predictions are more likely to be accurate in this case than the earlier ones50.