Protein structural homology is often strong in the absence of significant sequence homology. Structure-based fold recognition methods have been shown to be useful to recognize possible structural resemblances even at levels of non-recognizable sequence similarity. Threading techniques are able to predict the 3D structure of a query protein by pulling its amino acid sequence through the backbone of experimentally determined protein 3D structures Perifosine without relying on sequence similarity. Thus, these techniques can complement sequence-based methods in structural and functional VE-821 ATM/ATR inhibitor annotation of proteins. Detection of very remote homology by using these techniques has been successfully demonstrated in previous applications. In our previous work, we discovered a novel remote member of the chemokine family by applying fold recognition methods, the human chemokine CXCL17, which is the last member of the CXC chemokine family being so far identified. Chemokines are secreted signal proteins with significant impact on the function of the immune system and are important molecules in inflammatory responses. Some chemokines have also been shown to play a role in processes like angiogenesis, haematopoiesis, inhibition of HIV infection, tumor growth and apoptosis. Furthermore, they are very well suited for the development of small molecule inhibitors with strong therapeutic potential as they act through G protein�C coupled receptors. Chemokines share a conserved 3D structure, the so-called IL8- like chemokine fold, which is stabilized by cysteine residues forming intra-molecular disulfide bonds. Interestingly, the predicted IL8-like chemokine structure of CXCL17 revealed disulfide bonds in non-canonical regions in 3D structure but still maintaining an active fold. The low sequence similarity to other known members of the family and its cysteine patterns differing from those in known chemokines are the reasons why chemokine CXCL17 escaped annotation by standard sequence-based methods. Current standard techniques that can be applied to identify chemokines include the ChemoPred web server, a machine learning technique, Hidden Markov Models of SMART and Pfam and PROSITE��s chemokine profiles and patterns. However, the results that can be obtained by these methods are limited to the diversity of the training dataset, which originates almost exclusively from sequence-based approaches. Thus, these methods can hardly detect structural resemblance without detectable sequence similarity. The identification of CXCL17 by threading-based computational means motivated us to perform a more systematic search in the human proteome to discover other possible remote members of the chemokine family that, like CXCL17, might have been so far overlooked. Several threading techniques are available to detect structural resemblances in proteins, including the methodology we used to discover CXCL17.