
In this directory are input files, parameter files, and output files for 3 example runs of MotEvo
in its TFBS prediction mode.

Input files:
------------
input_ref.fna = 11 proximal promoter regions on chromosome 7 of the human genome.
input.fna = Multiple alignments with other mammals for the same 11 proximal promoter regions.

Weight matrix files:
wms_noweight = file with 3 example mammal weight matrices.
wms_weight = the same 4 WMs but now with different prior weights assigned.

Background model:
bg_order1 = probabilities for a bg-model of order 1 (single neighbor).

Parameter files:
params_ref = Parameters for running only on the human sequence, 
using a simple zeroth-order background model, and fixed prior.

params_ma = Parameters for the same as previous but now on the
multiple alignment, and printing log-likelihoods for each sequence

params_prior_bg = In this version MotEvo also estimates the priors and uses a first
order background model read from a file.


Example 1
---------
Here we run only on the human sequence. The priors are the same for all motifs and are fixed. 
A zeroth order background model with equal probabilities for each letter is used. Only sites 
with posterior probability at least 0.2 are printed out.

Command:
../../source/motevo input_ref.fna params_ref wms_noweight > report_ref

output files:
report_ref = This contains information on the mode in which MotEvo was run.
sites_ref = Predicted TFBSs.
priors_ref = Predicted priors and densities.

Example 2
---------
Here we run in the exact same way as in example 1, except that we now
use multiple alignments as input and print out log-likelihoods of each
sequence. We now use the UFE model, with length 8, and note also that
we print out all predicted UFE sites in the output files.

Command:
../../source/motevo input.fna params_ma wms_noweight > report_ma

output files:
report_ma = Information on the mode in which MotEvo was run.
sites_ma = Predicted TFBSs.
priors_ma = Predicted priors and densities.
loglikelihoods_ma = Log-likelihoods of each alignment under our model.

Example 3
---------
Here we give different priori weights to the input WMs (these are specified by the 'PW' lines in the WM file), 
we use expectation-maximization to fit all the priors, and we use a background model of order 1 that is
read from a file. In addition, we do now omit the UFE sites from the reported sites.

Command:
../../source/motevo input.fna params_prior_bg wms_weight > report_prior_bg

output files:
report_ma_priorbg1 = Information on the mode in which MotEvo was run.
sites_ma_priorbg1 = Predicted TFBSs.
priors_ma_priorbg1 = Predicted priors and densities.

Note that with prior updating only 1 Snail site remains over the 0.2 posterior cut-off at the end.

