Multiply two tf transforms converted to 4x4 matrices in python?

edit

The transformations.py included in tf provided the functions for the solution, they create 4x4 numpy nd.arrays which are matrix multiplied with dot():

(trans1, rot1) = tf.lookupTransform(l2, l1, t)
trans1_mat = tf.transformations.translation_matrix(trans1)
rot1_mat   = tf.transformations.quaternion_matrix(rot1)
mat1 = numpy.dot(trans1_mat, rot1_mat)

(trans2, rot2) = tf.lookupTransform(l4, l3, t)
trans2_mat = tf.transformations.translation_matrix(trans2)
rot2_mat    = tf.transformations.quaternion_matrix(rot2)
mat2 = numpy.dot(trans2_mat, rot2_mat)

mat3 = numpy.dot(mat1, mat2)
trans3 = tf.transformations.translation_from_matrix(mat3)
rot3 = tf.transformations.quaternion_from_matrix(mat3)

br = tf.TransformBroadcaster()
br.sendTransform(
  trans3,
  rot3,
  t,
  "target",
  "source");