-------------- Functional Association Matrix (FAM) ------------
-------------- Release Date : 30th July, 2014 ----------------
-------------- File created by : Ishita K Khan, Ph.D Student, Compute Science -----------------

Functional Association Score (FAM) score is computed between 2 GO terms as a contexual association between the terms in annotation database Uniprot.org. Briefly, the method to compute FAM score consists of the following steps -

1. Existing GO annotations for proteins are taken from Uniprot.org. 
2. All the annotations are propagated to the root of the ontology. 
3. For a pair of GO terms GO1 and GO2, the FAM score FAM(GO2 | GO1) is computed as the contexual association of GO1 and GO2 relative to the frequency of occurance of GO1 in the annotation database, using the following formula - 

FAM(GO2|GO1) = [ c(GO1, GO2) + epsilon ] / [ c(GO1)  mu * epsilon] 

Here, c(GO1, GO2) is the number of times GO1 and GO2 are used togather to annotate a protein, c(GO1) is the frequency of GO1 in annotation database, mu is the total number of unique GO terms considered in the associations, and epsilon is the pseudo-count, which is set to 0.01.

The link FAM Scores[Old Version] contains FAM scores computed in 2008 and have a total of 503207 FAM associations. The link FAM Scores[Updated Version] contains FAM scores computed in 2013 and have a total of 2936784 FAM associations. 

For reference, please cite - 

1. Troy Hawkins*, Meghana Chitale* & Daisuke Kihara (* equal contribution), Functional enrichment analyses and construction of functional similarity networks with high confidence function prediction by PFP., BMC Bioinformatics, 11:265.(2010).

2. Troy Hawkins, Stan Luban & Daisuke Kihara, Enhanced automated function prediction using distantly related sequences and contextual association by PFP." , Protein Sci, 15:1550-1556 (2006).