Virtual markers / dots: 

The domain data is arranged as an MxN matrix, where M is the number of frames.
Each row contains the 3D world positions(x,y,z) for the virtual markers in the
following order: top of head, neck, base of neck, upperback, lowerback, pelvis, upperarmR,
lowerarmR, handR, fingertipsR, upperarmL, lowerarmL, handL, fingertipsL,
upperlegR, lowerlegR, footR, toesR, tips of toesR, upperlegL, lowerlegL,
footL, toesL, tips of toesL

If we read the text file into a matrix called 'database', then database(1,1:3)
will contain the x,y,z positions of the "top-of-head" marker for frame #1. And
database(1,10:12) will contain the x,y,z positions of the "upperback" marker
for frame #1.

-----------------------------------------------------------------------------------------------------

Joint angles:

We also have the domain data as hierarchical joint angles. These are arranged
as an MxJ matrix, where M is the number of frames. Each row contains the joint angles(theta_x,theta_y,theta_z) for a given limb segment with respect to its parent in the following order: "pelvis", "lfemur", "ltibia", "lfoot", "ltoes", "rfemur", "rtibia", "rfoot", "rtoes", "lowerback", "upperback", "lclavicle", "lhumerus", "lradius", "lhand", "rclavicle", "rhumerus", "rradius", "rhand", "LowerNeck", "Neck"

If we read the text file into a matrix called 'jointdatabase', then jointdatabase(1,1:3)
will contain the x,y,z positions of the "pelvis" marker for frame #1 and
jointdatabase(1,4:6) will contain the world orientation of the "pelvis".
After that, we have the hierarchical joint angles, for example, jointdatabase(1,13:15) will contain the orientation of "lfoot"
with respect to "ltibia".