This is a bioinformatics project done at the Bromberg Lab at Rutgers University. It is a metagenomics experiment, separated into 6 steps: 1) Obtain a genome and break it up into n fragments of desired size range. 2) Translate these fragments directly to their corresponding peptide, further splitting fragments by stop codons, and filtering out those fragments with len < 11 3) Perform a PSI-BLAST on the fragments against a specified library 4) Make an 'outfile' of the results of the PSI-BLAST 5) Parse this outfile, looking for queries that hit a different gene with the same function (specified by COG family) 6) Mathematics and statistics (magic) These steps will return a graph that gives us a curve. We can then plot a 'test set' onto this graph, and based on the distances between each member of the set and the curve, it is possible to determine the function of that member, even if we do not know the COG family.