All evaluations in CAFASP-2 will be performed by Fully Automated programs, but the full responsibility of running the programs and present the results will lie entirely on each of the categories' coordinators . The procedures applied by these programs are delineated below. Participation in CAFASP-2 implies acceptance of these procedures and rules. The results of the automatic evaluation will be presented as they become available and will be independent of any other human assessment that may be carried out. All data used for the automated evaluation will also be available to all at all times.

In addition to the automated evaluation of CAFASP-2 predictions, two comparative analyses of the automated results filed to CAFASP-2 with those filed at CASP-4 may be carried out. These are described at the bottom of this page.

CAFASP-2 EVALUATION DESCRIPTIONS:

• Fold Recognition Evaluation Rules (Fischer)
• Secondary Structure Evaluation Rules (Rost)
• Homology Modeling Evaluation Rules (Dunbrack).
• Ab initio Evaluation Rules (Ortiz)
• Contacts Prediction Evaluation Rules (Valencia, to appear soon) .

CAFASP-CASP PERFORMANCE COMPARISONS:

The two comparative analyses of the automated results filed to CAFASP-2 with those filed at CASP-4 that will be carried out are:

• 1. CAFASP-CASP automated evaluation comparison. After the CASP-4 predictions become available to all, we may apply to them the exact same procedures as those described above.
• 2. CAFASP-CASP human assessor comparison. The CASP-4 assessor of each category may apply his/her assessing procedures to both the CASP-4 and CAFASP-2 predictions.

It is obvious that a CAFASP-CASP performance comparison is not 100% fair for several reasons:

• The CAFASP-2 predictions are filed months before the CASP-4 predictions are filed. Thus, it is likely that for some cases, the availability of more sequence or structure entries in the databases may make a particular target easier to predict.
• The manual casp4 predictions can make use of the CAFASP results (but not vice-versa).