This server address the threading alignment problem directly in an exhaustive fashion. We split the protein sequence recursively and align several combinations to identify the domains. The final selection is based on the threading score. This is the input to structure prediction and refinement pipelines.