Features of TEXTAL(TM) in PHENIX

Summary of TEXTAL(TM) in PHENIX Command line Interface Interface to TEXTAL(TM) in the PHENIX GUI TEXTAL(TM) manual

The latest version of TEXTAL(TM) can build the protein models with structure factors of resolution greater than or equal to 3 angstrom automatically. The goal of automated model building has been furthered by eliminating the need for the user to generate an electron density map at 2.8 angstrom from their structure factors. The current process of automated model-building potentially saves the crystallographer a great deal of time and allows them to go directly from structure factors to a TEXTAL(TM) built model. TEXTAL(TM) still maintains the capability to build protein models with an electron density map as the input. A recent addition to TEXTAL(TM) Findmol makes it possible to automatically center the electron desnity map over a single macromolecule eliminating the previous need for an electron density map over a single molecule as input.

TEXTAL(TM) uses the priniciples of pattern recognition to build a protein model. The pattern recognition algorithm looks for patterns similar to the local region from a database of maps for proteins whose structures have already been solved. An efficient database search is carried out using Feature Extraction . There are 3 main phases to TEXTAL(TM) : CAPRA, LOOKUP, and post-processing. CAPRA (Cα Pattern Recogniiton Algorithm) models the protein backbone. Typically, the protein bacbone returned by CAPRA is comprised of multiple chains. LOOKUP models the side-chains of the protein. While the traditional input to LOOKUP is the CAPRA built protein backbone, LOOKUP has the ability to model side-chains for user input protein backbone. LOOKUP is potentially time consuming and takes 1-2 hours for large proteins. The post-processing routines include

Sequence Alignment: Maps individual chains onto the protein sequence.
Real-space Refinement: Fits side-chains better to density and adjusts bond distances and angles across peptide connections between neighboring residues (modeled independently by LOOKUP).

TEXTAL(TM) is now available both in the PHENIX GUI and also on the command line. The GUI supports model building from structure factors provided only in MTZ format, but it is possible to input other (hkl, sca, etc.) reflection file formats from the command-line.