The files exonrank2_sorted.html and exonrank2_unsorted.html in this
directory list the results of our analysis to detect exons whose
sitewise dn/ds (omega) values do not conform to the background
dsitribution derived from all the proposed protein-coding parts of the
Encode regions.

Note that each row corresponds to one proposed exon.

In file exonrank2_sorted.html, the exons are sorted according to their
'oddness' (most strange exons first).

In file exonrank2_unsorted.html, the exons are ordered according to
target region and exon_id (see below).

In each file, the columns are:

  exon_id:  gene name + exon's Encode coordinates + reading frame (links
            to plot of sitewise dn/ds values)

  plot_coords:  exon's first and last nucleotide positions in our own
                gene position labelling system (corresponds to positions
                in the sitewise dn/ds plots)

  region:  Encode target region

  splicing:  indicates constitutive or alternatively spliced or
             information not available (const/alt/nA)

  num_codons:  no. of codons

  pos_sites:  no. of sites with significant evidence of positive selection
              (dn/ds>1)

  weimean_omega:  weighted mean dn/ds

  best_scenario, worst_scenario, avg_score:
    We use two measures of 'oddness' for an exon, one (best_scenario)
    based on the optimistic scenario that all its sitewise omegas
    coincide with the lower bounds of their confidence intervals and one
    (worst_scenario) based on the pessimistic scenario that all its
    sitewise omegas coincide with the upper bounds of their confidence
    intervals.  We use the average of these two values as a sorting
    criterio.  More detailed classification is possible using the rules:

An exon should be considered as good if its worst_scenario score is low.
An exon should be considered as bad if its best_scenario score is high.