Showing Revision 5 created 01/15/2019 by Paulo Henrique Ferreira.

1. Title:
2. Description:

3. 3824 0:04 - 0:08
   Understanding the causal origins and
   mechanistic principles for the behavior

   ¶

4. 8294 0:08 - 0:13
   of an evolving system is one of the major
   challenges of our time.
5. 12794 0:13 - 0:18
   For example, how a protein may fold to
   become a functional piece in your body
6. 18147 0:18 - 0:22
   or how a drug may help against
   some disease.
7. 21680 0:22 - 0:28
   A causal system can be described by an
   algorithmic model evolving over time.
8. 27511 0:28 - 0:30
   The length of the shortest computer model
   is called
9. 30468 0:30 - 0:35
   the algorithmic information content of
   the system.
10. 34514 0:35 - 0:41
    That is how much computer code is needed
    to reproduce the object itself.
11. 40675 0:41 - 0:47
    The shorter the description, the more
    likely the system is causally generated.
12. 46538 0:47 - 0:51
    The longer its description, the less
    likely it is to be causally generated.
13. 51413 0:51 - 0:55
    A dynamic system can usually be
    represented
14. 54607 0:55 - 0:57
    as a network of interacting elements,
15. 57181 0:57 - 1:04
    such as interacting cells or interacting
    genes turning on and off other genes.
16. 64172 1:04 - 1:09
    What we want is to figure out the first
    principles driving a system,
17. 69000 1:09 - 1:13
    such as a genetic network representing a
    cell
18. 72593 1:13 - 1:18
    or the causes for the cell to behave in
    one way or another.
19. 78250 1:18 - 1:23
    Power approach consistis in finding a set
    of algorithmic models that can explain
20. 83319 1:23 - 1:26
    the structure of the system.
21. 86030 1:26 - 1:31
    When the system is just a random process
    and, therefore, not causally generated
22. 91253 1:31 - 1:37
    it can only be represented by a long
    descriptive model.
23. 96684 1:37 - 1:42
    However, systems rich in causal content
    can be represented by a short model,
24. 102177 1:42 - 1:46
    because they have a generating mechanism
    that evolves
25. 105631 1:46 - 1:50
    into any observable state of the system.
26. 109912 1:50 - 1:53
    To determine the causal content of
    an evolving system,
27. 113070 1:53 - 1:55
    we peform perturbations.
28. 114739 1:55 - 2:01
    In a network, for example, a perturbation
    can be deleting a node or deleting a link.
29. 121377 2:01 - 2:06
    We examine the effects of that
    perturbation, and evaluate how much the
30. 126171 2:06 - 2:10
    system became more or less random.
31. 129504 2:10 - 2:12
    In a random system, for example,
32. 131901 2:12 - 2:17
    a change will not have a major impact,
    because no part of the system can explain
33. 137269 2:17 - 2:23
    any other part of the system, and so
    the perturbation goes unnoticed.
34. 142747 2:23 - 2:29
    We can thus safely say that these
    systems cannot be reprogrammed.
35. 149349 2:29 - 2:33
    But, in non-random causally generated
    systems,
36. 153299 2:33 - 2:37
    some changes will render them
    unrecognizable,
37. 156915 2:37 - 2:43
    causal interventions in these systems make
    them reprogrammable.
38. 162689 2:43 - 2:49
    Moreover, if we remove an element and the
    system gets further away from randomness,
39. 168840 2:49 - 2:55
    we can conclude that the element is not
    part of the algorithmic content,
40. 174524 2:55 - 2:59
    and is unlikely part of the causal
    generating mechanism of the network.
41. 179167 2:59 - 3:05
    It is thus likely to be noise or part of
    another system.
42. 184577 3:05 - 3:10
    Alternatively, if we remove an element and
    the system moves towards randomness,
43. 190000 3:10 - 3:14
    it means that the element is part of the
    causal model that explains
44. 194088 3:14 - 3:17
    other parts of the system's evolution.
45. 196592 3:17 - 3:22
    But, if we remove an element and the
    system does not approach either
46. 201579 3:22 - 3:27
    randomness nor simplicity, the deleted
    element is non-essential in the
47. 206840 3:27 - 3:30
    explanation of the system and likely an
    element produced by
48. 210354 3:30 - 3:33
    the normal course of its evolution.
49. 213230 3:33 - 3:38
    Returning to gene regulation, what we do
    is to apply this concept
50. 217977 3:38 - 3:43
    and evaluate the contribution of
    every gene and gene interaction to the
51. 222990 3:43 - 3:48
    original network, ranking the elements by
    causal contribution.
52. 228798 3:49 - 3:52
    When this ranking is biologically
    interpreted,
53. 231622 3:52 - 3:56
    these tools have demonstrated the ability
    to pinpoint markers of
54. 236087 3:56 - 4:01
    cell function, cell differentiation, and
    cell fate.
55. 240678 4:01 - 4:07
    This shows how these tools can properly
    profile systems elements, and steer,
56. 246535 4:07 - 4:12
    and reprogram systems, such as biological
    cells.
57. 251693 4:12 - 4:16
    If these systems have some elements that
    move the network towards randomness,
58. 256172 4:16 - 4:21
    and other elements that move the network
    towards simplicity,
59. 260530 4:21 - 4:25
    we say these systems are more
    reprogrammable.
60. 265093 4:25 - 4:31
    But, if a system can only move in one
    direction, then it is less reprogrammable.
61. 270540 4:31 - 4:35
    Stem cells will be able only to move
    towards a single direction
62. 274654 4:35 - 4:38
    towards a differentiated cell,
63. 277942 4:38 - 4:41
    while completely differentiated cells will
    be able only to move
64. 281240 4:41 - 4:45
    to the opposite direction.
65. 284620 4:45 - 4:49
    This causal calculus is better equipped to
    tackle the general challenge of
66. 289146 4:49 - 4:53
    causal discovery and science than more
    traditional tools,
67. 292682 4:53 - 4:57
    and is helping scientists better
    understand, and even steer,
68. 297157 4:57 - 5:00
    biological and synthetic systems.