Biology Reference
In-Depth Information
44. To give a sense of their ubiquity, Weinstein estimates that there were
about 4,000 scientifi c papers published between 1998 and 2008 that included
heat maps. Moreover, the introduction of the use of heat maps for microarray
analysis (Eisen et al., “Cluster Analysis,” discussed below) was, in July 2008,
the third most cited article in
Proceedings of the National Academy of Sciences
USA
. See Weinstein, “Postgenomic Visual Icon.”
45. The simplest way to do this is to use a distance metric based on the Py-
thagorean theorem extended to higher-dimensional spaces. For instance, if one
measured the levels of four proteins in four patients, the “distance” between
the measured levels in any two patients would be given by
(
2
2
2
2
aa
−+−+−+−
)
(
bb
)
(
cc
)
(
dd
)
,
where
i
and
j
are the two patients and
a
,
b
,
c
,
d
are the four proteins. Once all
the distance pairs have been measured in this way, the algorithm would cluster
the two patients with the smallest distance into a node on the tree. The posi-
tion of the node is given by an average of
i
and
j
. The process is then repeated
using the remaining two patients and the new node, again searching for the
shortest distance. The process is repeated until all the patients are joined into a
single tree.
46. Weinstein reports that in one case, a data set from molecular pharma-
cology was reduced from a 8.3-meter scroll to a one-page heat map with 4,000
lines. Weinstein, “Postgenomic Visual Icon,” 1773.
47. Eisen et al., “Cluster Analysis.”
48. Sneath, “The Application of Computers.”
49. Hagen, “The Introduction of Computers.”
50. Wilkinson and Friendly, “History of the Cluster Heat Map.” The Paris
data are taken from Loua,
Atlas statistique
.
51. Weinstein, “Postgenomic Visual Icon,” 1772.
52. Alternative splicing of mRNA transcripts is described in chapter 2. For
a detailed but introductory account of alternative splicing, see Matlin et al.,
“Understanding Alternative Splicing.”
53. The amount of data produced was equivalent to about four com-
plete human genomes. The production of these data was made possible by
the sequencing speed of the new machines such as those made by Illumina.
See www.illumina.com and Bentley et al., “Accurate Whole Human Genome
Sequencing.”
54. Wang et al., “Alternative Isoform Regulation.”
55. On “mechanical objectivity,” see Daston and Galison,
Objectivity
.
56. Fry, “Computational Information Design.”
57. Fry, “Computational Information Design,” 11.
58. See http://www. benfry.com/salaryper/.
59. To see many of these tools in action, go to http://benfry.com/projects/.
60. Fry, “Computational Information Design,” 82-83.
61. Fry, “Computational Information Design,” 89-90.
62. Fry, “Computational Information Design,” 110.
63. Reas and Fry,
Processing
. See also www.processing.org.
64. Casey Reas, quoted in Fry, “Computational Information Design,” 126.
i
j
i
j
i
j
i
j
