Assignment of function and quantification of placements

[Robaina]

Two ideas:

1. Assignment of function could be done along with taxonomy by assigning a "fake" taxonomy to reference sequences contained in invalid clusters (such as those composed of paralogs). The taxonomy assignment algorithm would then tag placed sequences within invalid clusters as invalid based on the fake taxonomy. Sequences placed in proper clusters would be assigned the expected function according to the tree
2. Quantification of placed sequences is done according to the new categories created by assigning function and taxonomy to placed sequences

[Robaina]

Ok, the above has been implemented

Two ideas:

1. Assignment of function could be done along with taxonomy by assigning a "fake" taxonomy to reference sequences contained in invalid clusters (such as those composed of paralogs). The taxonomy assignment algorithm would then tag placed sequences within invalid clusters as invalid based on the fake taxonomy. Sequences placed in proper clusters would be assigned the expected function according to the tree
2. Quantification of placed sequences is done according to the new categories created by assigning function and taxonomy to placed sequences

[Robaina]

Now, we still have to handle placement quality scores, we have two types:

1. Reference tree cluster quality score: This is already implemented
2. Query placement quality score: This score measures the "goodness" of placement of a particular query sequence. The score could summarize: the distance of the placement point to the main branch, the average sequence identity of the query w.r.t. the rest of the cluster, and/or some other parameter.

Note that the jplace output by epa-ng assigns a probability to each placement (a query sequence may have more than one placement, in which case, we are currently taking the placement with the highest probabilty, per gappa examine graft)