Critical Assessment of Fully Automated Structure Prediction


DP is a new category in CAFASP4: Domain Prediction

To evaluate the capabilities of current programs to identify domains in protein sequences, CAFASP4 will extend its scope by introducing a new category: Domain Prediction (DP).

See the list of participating DPs.

Domain Prediction Servers will receive as input the amino acid of the target sequence and will return the predicted domain(s) in the format below, illustrated on the hypothetical target T2001 ADPPRSKVVNLMSWT:

1 A 1
2 D 1
3 P 1
4 P 1
5 R 2
6 S 2
7 K 2
8 V 2
9 V 2
10 N -
11 L -
12 M 1
13 S 1
14 W 1
15 T -

The first column should contain the residue numbers, starting from number 1. The second column should contain the one-letter residue from the sequence. The third column is either the predicted domain number or a "-", if nothing is predicted.
Notice that the above example means: the prediction identifies 2 domains, the first being non-contiguous, and residues 10, 11 and 15 have no prediction.
Optionally, servers can add additional columns, e.g. to specify per-residue confidence scores.

To participate in CAFASP4 DP, a DP server must accept queries by email in the following format:

CAFASP DP target T2001


The DP server should reply by email to the sender with the prediction in the format described above. Please note that we will only evaluate predictions that have the exact above format.

To enter your DP in CAFASP contact Daniel Fischer, and please provide the email address to which queries should be sent.