Ebook The cell language theory - Connecting mind and matter: Part 2

Part 2 book “The cell language theory – Connecting mind and matter” has contents: Applications of the cell language theory to biomedical sciences, the universality of the planckian distribution equation, the universality of the irreducible triadic relation, the philosophical implications of the cell language theory, conclusions.

Chapter 7

Applications of the Cell Language Theory to Biomedical Sciences

Most, if not all, human diseases, both somatic and mental, can be said to

arise from miscommunication and disregulation of metabolism within

individual cells (i.e., intracellular semiosis) or between cells (i.e.,

intercel-lular semiosis) in the human body Hence, the cell language theory and

medical sciences are intimately related

The cell language theory is one of the four major theoretical building

blocks underlying the theoretical model of the living cell known as the

Bhopalator discussed in Chapter 3 The four components of the Bhopalator,

i.e., cell language, cell force, intracellular dissipative structures (IDSs),

and conformons, are depicted in Figure 7.1 as the four nodes of a

body-centered tetrahedron (BCT) whose center is occupied by the cell model

One unique feature of the tetrahedron is that its four nodes are all

equiva-lent and in simultaneous contact with one another, which is used in Figure 7.1

as a topological means to symbolize the essentiality and the

interconnect-edness of the four theoretical components of the living cell In other

words, these four theoretical building blocks constitute the irreducible

tetrad of the cell (ITC) The principle of ITC implies that the cell structure

and function cannot be completely accounted for without implicating all

of the four theoretical components, although, at any given time, only one

or two of them may be prescinded (i.e., selected out or highlighted for

Figure 7.1 Cell language as one of the four major building blocks of the theoretical

model of the living cell, the Bhopalator The geometric figure employed here is called

the BCT (Body-Centered Tetrahedron) (see Figure 10.15) that has been found to provide

a useful template for modeling many processes and structures in the Universe (see

emphasis, temporarily ignoring the rest of the components, for the

conveni-ence of thought) As we attempt to apply the cell language theory in this

chapter to solve practical problems in biomedical sciences, our emphasis will

be placed on the intercellular or intracellular communications (or semiosis)

mediated by the cell language, but this does not mean that the other three

theoretical components are not involved in one way or another

7.1 The Need for a New Paradigm in Biomedical Sciences

H H Heng, the author of Debating Cancer [297], recently stated that:

The explosion of genomic information has generated both excitement

and confusion The paradox of knowing more about cancer’s genetic

landscape yet understanding less of its common molecular basis

repre-sents such an example It was believed or hoped by many that the

cancer genome sequencing project would once and for all solve the

mystery of cancer, without anticipating that the powerful technology

would further add to the unmanageable complexity of the picture

Following the high hopes of the development and utilization of various

large scale -omics technologies, the long expected clear-cut

under-standing of cancer is actually fading away… What is the real

prob-lem? Not enough molecular data yet? No suitable model for data

analyses? Or on an even more serious note, has there been a wrong

Of the three possible explanations for the “cancer paradox” that Heng

is conceptualizing, I think that the last possibility is the most likely

expla-nation, i.e., a wrong conceptual framework for not only cancer research,

but also for the biomedical science research and education, in general.

7.1.1 The Inefficiency of the Current Methods of Drug Development

One evidence for the “wrong conceptual framework” of the contemporary

biomedical science, I think, is provided by the inefficiency of the current

drug development research According to Bain & Company [298], the cost

of developing a new drug is estimated to be $1.7 billion and it takes 12–16

years to complete a drug development process from the compound

discov-ery stage to marketing The overall attrition rate for developing a drug is

calculated to be 10,000:1 According to another survey, the United States

invested a total of $25 billion in 2000 on the research and development for

pharmaceuticals and produced only 11 new drugs on the market in that

year, costing the US pharmaceutical industry $2.3 billion per new drug In

addition, once a drug is approved by the FDA, the success rate of drug

treatment is only 30–60% [299]: Only about 50% of the patients treated

with drugs respond favorably

7.1.2 Precision Medicine

In his State of the Union Address on January 30, 2015, President Obama

launched the Precision Medicine Initiative with the following statement:

Doctors have always recognized that every patient is unique, and

doc-tors have always tried to tailor their treatments as best they can to

individuals You can match a blood transfusion to a blood type — that

was an important discovery What if matching a cancer cure to our

genetic code was just as easy, just as standard? What if figuring out the

right dose of medicine was as simple as taking our temperature? (7.2)

The cell language and associated biological theories described in this

book suggest one possible strategy for implementing the Mission

Statement (Figure 7.2(b)) of the Precision Medicine Initiative, as briefly

summarized in Figure 7.2(c) and 7.2(d) and in the figure legend

Figure 7.2 A possible strategy for implementing the Precision Medicine is based on: (i)

the cell language theory (i.e., the Bhopalator model of the cell), (ii) the microarray

tech-nology, and (iii) PDE (as a quantitative method for classifying the normal and disease

states of the human body (see Section 7.3 for specific examples), the theoretical model of

which being the Piscatawaytor (formulated in 1991).

President Obama Retrieved from https://www.whitehouse.gov/precision-medicine on

01/04/2016.

The Mission Statement:

“To enable a new era of medicine through research, technology, and policies that empower patients, researchers, and providers to

work together toward development of individualized treatments.”

Efficient Drug Development Precision Medicine

The proposed strategy of implementing the Precision Medicine

Initiative (PMI) is based on three components: (i) the cell language-based

model of the living cell, the Bhopalator, and the human body, the

Piscatawaytor (see Section 3.2.20), (ii) the microarray technique for

measuring mRNA levels in cells and tissues (see Section 7.2), and (iii) the

Planckian distribution equation (PDE) (described in Chapter 8) that

intro-duces a new quantitative method for classifying long-tailed histograms of

mRNA levels measured from both normal and diseased cells and tissues

Since the microarray technique plays a fundamental role not only in the

proposed strategy for implementing PMI, but also in possibly ushering in

a paradigm shift in cell biology and medicine, this method and its

implica-tions in biology are discussed in some detail in the following secimplica-tions

7.2 Ribonoscopy

The term “ribonoscopy” was coined in 2012 [25] to indicate the scientific

study of mRNA levels in living cells and tissues measured with DNA

microarrays, in analogy to spectroscopy which is the study of optical

spectra of atoms and molecules using spectrometers [300] “Ribonoscopy”

is an experimental method by which

We can study living cells using RNA molecules and their copy number

variations as molecular reporters of intracellular events (7.3)

7.2.1 DNA Microarrays

A microarray consists of a microscopic slide (or its equivalent), about

2 cm by 2 cm in dimension, divided into, typically, 10,000 squares or

spots, to each of which is covalently attached a fragment of DNA (i.e.,

cDNA, or oligonucleotides) that is complementary to a stretch of the

genome encoding an RNA molecule Thus, using one microarray, it is

possible to measure simultaneously the levels of 10,000 RNA molecules

or more in a biological sample Before the development of the microarray

technique, it was possible to study only a small number of RNA molecules

at a time The experimental procedures involved in DNA microarray

measurements are schematically summarized in Figure 7.3 and its legend

A typical microarray experiment implicates the following steps:

Figure 7.3 The microarray experiment and typical results (a) The mRNA isolated from

a biological sample is transformed into complementary DNA (cDNA) using reverse

tran-scriptase and labeled nucleotides which is then hybridized with the probe DNA previously

attached to the microarray surface Image reproduced from [301, 302] (b) In two-color or

two-channel microarray experiments, complementary DNA molecules are prepared from

two samples to be compared, e.g., cancer vs normal cells, with differential labeling The

fluorescent dyes commonly used for cDNA labeling include Cy3 (cyanine dye 3), which

fluoresces at 570 nm (corresponding to the green color), and Cy5 which fluoresces at

670 nm (corresponding to the red color) The two Cy-labeled cDNA samples are mixed

and hybridized to a single microarray that is then scanned in a microarray scanner to

visu-alize fluorescence of the two fluorophores by exciting with a laser beam or 570 or 670 nm

wavelength The relative intensities of each fluorophore are analyzed based on a

ratio-based method [303] after proper normalization [304] to identify up-regulated and

(c)

(d)

1 Isolate mRNA from broken cells.

2 Synthesize fluorescently labeled cDNA from mRNA using reverse

transcriptase and fluorescent nucleotides

3 Prepare a microarray either with DNA fragments or oligonucleotides

synthesized on the microarray surface

4 Pour the fluorescently labeled cDNA preparations over the

microar-ray surface to effect hybridization Wash off excess debris

5 Measure fluorescently labeled cDNA using a computer-assisted

fluo-rescence microscope

6 The final result is a table of numbers, each number registering the

fluorescent intensity which is in turn assumed to be proportional to

the concentration of cDNA (and ultimately mRNA in the cell) located

at row x and column y, the row indicating the identity of genes, and

y the conditions under which the mRNA levels are measured (see

Table 7.2)

Another example of microarray measurements of RNA is shown in

Figure 7.4 The term RNA here refers to not only mRNA, but all other

forms of RNA including RNA complementary to the introns, promoters,

ribosomal RNA, small interfering RNA, and non-coding RNA The data

in Figure 7.4 were measured by Garcia-Martinez et al [315] from

bud-ding yeast Saccharomyces cerevisiae undergoing glucose–galactose shift

at six time points: 0, 5, 120, 360, 450, and 850 min after the nutritional

shift Each data point is the average of three measurements The overall

quality of the kinetic data, as evident in the smooth and coherent trajectory

exhibited by each gene, increases our confidence in the microarray

experi-mental method

Figure 7.3 down-regulated mRNA levels The image is adopted from [305] (c) The mRNA

fold changes (y-axis) in breast tumor tissues of 20 patients (x-axis) before (BE) treatment

relative to control (N) Each profile represents the mRNA level changes encoded by one

gene The figure contains 50 genes out of ~5000 genes analyzed (d) The mRNA fold

changes of tumor in 20 breast cancer patients before (BE) and after (AF) treating with

doxo-rubicin for 16 weeks RAT2N indicates mRNA level ratio with red channel (or channel 2)

normalized [306].

The advent of the microarray technique in molecular biology in the

mid-1990s [307–313] marks an important turning point in the history of

cell biology, comparable to the discovery of DNA double helix in 1953

Although there remain many challenging problems, both methodological

[314] and biological [301], this novel technology possesses a great

poten-tial to make fundamental contributions to advancing our basic knowledge

about the workings of the living cell, with important consequences in

medicine, biotechnology, and pharmaceutical industry

7.2.2 The Microarray Data Interpretation Problem

It is unfortunate that, from the beginning of the microarray era, leaders in

the field have created the impression that the microarray technique allows

Figure 7.4 RNA dissipative structures (or dissipatons) encoding glycolytic enzymes The

intracellular levels of the RNA molecules encoding glycolytic enzymes are measured in

budding yeast using DNA arrays by a Garcia-Martinez et al in Valencia [315] at t = 0, 5,

120, 360, 450, and 850 min after switching glucose with galactose Of the 13 trajectories

shown, the one labeled YCL040W (light blue) exhibits an unusual behavior of increasing

(rather than decreasing) between 5 and 120 min One possible explanation for this

observa-tion is that the degradaobserva-tion of the YCL040W transcript is selectively suppressed following

the nutritional shift.

Dissipative Structures (Glycolysis)

0 50 100 150 200 250

biologists to measure rates of gene expressions (denoted as TR,

transcrip-tion rates [315]) by measuring mRNA levels (denoted as TL, transcript

levels [315]) In other words, they have created the scientific atmosphere

in which it is deemed legitimate to accept a simple one-to-one

correspond-ence between TL and TR The following quotations reflect such a lax

attitude in the microarray field (emphasis is mine):

“… Microarrays prepared by high-speed robotic printing of

comple-mentary DNAs on glass were used for quantitative expression

“Oligonucleotide arrays can provide a broad picture of the state of the

cell, by monitoring the expression level of thousands of genes at the

“… DNA microarrays, permits the simultaneous monitoring of

These statements would be correct if the term “genes” (in italics) were

replaced by “mRNA levels” or “transcripts” In other words, workers in this

field routinely conflate “genes” with “gene transcripts” and Transcription

Rate (TR) with Transcript Level (TL), leading to numerous false-positive

and false-negative conclusions in interpreting microarray data Most

inves-tigators in the field seem to think that there is no harm in using the terms

“gene expression” and “mRNA levels” interchangeably, but the

investiga-tions by Garcia-Martinez et al [315] and Fan et al [316] have now clearly

demonstrated that the mixing of these two terms can lead to erroneous

conclusions [317–319]

Because of the experimental difficulties involved in measuring TR, it

was not until 2004 that J Perez-Ortin and his colleagues in Valencia, Spain,

succeeded in measuring both the TR and TL values simultaneously of the

whole genome of budding yeast subjected to glucose–galactose shift [320–

322] It is well known that when budding yeast cells are deprived of

glu-cose, they undergo a profound metabolic transition from fermentation

(converting glucose into ethanol) to respiration (converting ethanol to

car-bon dioxide and water) known as the diauxic shift [322] When these TR

values are plotted against the TL values, highly nonlinear trajectories were

obtained as shown in Figure 7.5 Previously investigators routinely assumed

that TR would be a simple linear function of TL, but as can be seen here,

TR is clearly not linearly related to TL in about half of the time (The

com-ponents of the TL–TR trajectories that are parallel to a straight line with a

slope of about 1 indicate linear correlations between TL and TR.)

Experimental evidence indicates that TL is determined by the balance

of two opposing processes — the transcription of genes into RNA or

mRNA (i.e., TR) and the degradation of mRNA into shorter fragments

(whose rate is denoted as TD, transcript degradation rates) [548], so that

the following relation holds:

where a, b, and c are the parameters whose magnitude may or may not

depend on individual mRNA nucleotide sequences If we assume that a

Figure 7.5 Plots of fold changes in TR and TL of budding yeast during metabolic

transi-tions caused by glucose–galactose shift These four examples (for mRNA molecules

encoded in genes #1, #3, #10, and #19) were chosen randomly out of the 5184 mRNA

molecules investigated by Perez-Ortin and his coworkers [315] Fold change in TL,

denoted by fTL, is defined as the ratio of TL at time t over the TL at t = 0, i.e., fTL = TL/

TL0 Each plot shows the results of six measurements at t = 0, 5, 120, 360, 450, and 850

min after glucose was replaced with galactose in the growth medium.

fTL-fTR Plot 1

0 0.2 0.4 0.6 0.8 1 1.2

6 6

and c are constant for the yeast genome and b is a function of individual

mRNA molecules (reflecting the peculiarities of the experimental method

for measuring TR, known as the nuclear run-on technique [315]), then Eq

(7.7) can be converted into

where A = b/a and B = c/a and “fX” indicates “fold changes in X” as

defined in the legend to Figure 7.5 Integrating Eq (7.8) leads to

fTL = ∫[A(fTR) - B(fTD)]dt (7.9)

We can draw two important conclusions from Eq (7.9):

(1) Since there are three variables in Eq (7.9), it is impossible to

deter-mine any one of them without also measuring one of the remaining

two For example, it would be impossible to determine A(fTR) by

measuring fTL alone (because of the B(fTD) term), contrary to

what has been routinely assumed in the field of microarray data

analysis, and

(2) Since there are at least three possibilities for the direction of changes

in d(fTL)/dt in Eq (7.8) — increase (+), no change (0), or decrease

(-) — and, for each one of which, there are again three possible

mechanisms for the term [A(fTR) - B(fTD)] to be (+), (0), or (-) [25,

Table 12.4], there are nine possible mechanisms for regulating

d(fTL)/dt and hence the TL values [273, 317].

Each of the nine possible mechanisms inferred above is associated

with a unique RNA turnover mechanism involving a system of enzymes

(e.g., RNA polymerase, ribonucleases, other regulatory factors), and hence

it is logical to refer to it as an RNA turnover module or simply RNA

mod-ules [273, 317] It should be pointed out that RNA modmod-ules invoked here

are examples of IDSs (see Sections 6.1.2 and 6.1.3), since they are not

permanent equilibrium structures such as RNA polymerases and

electron-transfer complexes but are transient ones that are called into action (or

excited or activated) by appropriate signals when needed and dissolve into

their components when their biological function is accomplished, very

similar to what Norris et al referred to as “hyperstructures” [69] Related

concepts are also discussed by Srere (“metabolons” [323]), Hartwell et al

(“modules” [190]), and Lehn (“supramolecular chemistry” [324])

The rich information contained in the TR and TL data measured by

Garcia-Martinez et al [315] can be more fully displayed in a

three-dimensional space consisting of the TR, TD, and TL axes (Figure 7.6) The

TD data were calculated from fTL and fTR data using Eq (7.8) For this

purpose, the dfTL/dt at any time point was computed by differentiating the

approximate TL function derived from TL data by an nth-degree polynomial

fitting procedure, where n is the number of measuring time points, i.e., 6.

One of the most striking features of the TR–TD–TL plots is that,

despite major changes in the TR and TD values, the TL values often

remain relatively constant This may suggest that, during the metabolic

perturbations caused by glucose–galactose shift, the yeast cell manages to

Figure 7.6 The three-dimensional plots sowing the dependence of the mRNA levels (TL)

on the rates of transcription (TR) (denoted as tr) and mRNA degradation (TD) (denoted as

v3) The vertical lines indicate the TL values plotted on the z-axis Each plot shows the

identity of the gene encoding the mRNA under observation These mRNA molecules

(coded by genes 1, 5, 6, and 8) are arbitrarily selected out of about 6000 mRNA molecules

investigated in [315] (I thank Drs Sunil Dhar and Robert Miura, both of NJIT, for their

help in preparing the plots shown in this figure.)

maintain mRNA levels constant as long as possible, despite the fact that

TR and TD undergo large changes Alternatively, it may be concluded

that, during the glucose–galactose shift, budding yeast cells regulate TR

interpretation as the “RNA homeostasis” or better, RNA homeodynamic

dynamic patterns of the changes in intracellular components, including

steady-state patterns Thus defined, homeodynamics includes homeostasis

as one of its species.) Similar phenomenon has been observed with respect

to the intracellular levels of ATP under a wide variety of cell metabolic

conditions [325]: i.e., the intracellular ATP levels remain relatively

con-stant despite great changes in the rates of ATP synthesis and utilization

One of the universal features of the dynamics of TL in the TL–TR–TD

space is the turning point occurring at around 120 min after the glucose–

galactose shift This is believed to be due to the metabolic patterns in

budding yeast switching from fermentation to respiration Therefore, we

can divide the trajectory of TL into two parts — one before and the other

after the turning point The trajectory before the turning point will be

referred to as the F (from fermentation) phase and that after the turning

point as the R (from respiration) phase The angle that the F and R phases

make at the turning point (to be called the “FR angle”) can be used as a

quantitative measure of the reversibility of the control mechanisms of

RNA metabolism in budding yeast: The smaller the FR angle, the more

reversible is the control mechanism of RNA metabolism (or the larger the

FR angle, the more irreversible is the control mechanism) Evidently, the

dynamics of the TL trajectory associated with gene 1 shows an almost

zero FR angle, whereas that associated with gene 6 exhibits an FR angle

close to 90° The reason for such differential behaviors exhibited by FR

angles is not yet clear

7.2.3 Ribonoscopy is to Cell Biology What Spectroscopy

is to Atomic Physics

Figure 7.7, which represents the states of gene expressions along

chromo-somes, shows striking, although superficial, similarity with atomic

absorption spectra such as shown in Figure 7.8 Figure 7.7 is about the

Figure 7.7 The C neoformans genome with each chromosome represented as a colored

bar Genomic features are pseudocolored, from red (high density) to deep blue (low

den-sity) These include the density of genes, transposons, expressed sequence tags (ESTs),

and predicted single-nucleotide polymorphisms (SNPs) [326].

Figure 7.8 The atomic spectra of the hydrogen atom (1) The hydrogen atom absorption

lines detected in the light from Zeta Tauri (2) The same absorption lines observed in the

light from another star, 11 Camelopaadlis [73, p 472].

locations and abundances of genes and related structures along the

chromosomes of the unicellular organism, Cryptococcus neoformans

[326] In contrast, Figure 7.8 shows the wavenumbers (i.e., the number of

waves per cm) of light absorbed when the electron in the hydrogen atom

undergoes transitions from one energy level to another [73, 327]

Figure 7.7 is about the cell and Figure 7.9 is about the atom, but they both

reflect the probabilities of some events occurring along appropriate

structural coordinates in each system

It may be useful to consider what may be referred to as the “gene

expression activity spectrum (GEAS)” which consists of the addresses

or locations of all the genes along chromosomes indicated on the x-axis

and the corresponding rates of gene expression (i.e., TR) along the

y-axis For the human genome, the GEAS would look very much like

Figure 7.7, only with a larger set of lines, approximately 1,000 per

chro-mosome, with varying heights reflecting different rates of

correspond-ing transcription

If the qualitative comparison given above turns out to be valid, cell

biologists might learn some useful lessons from the history of atomic

physics For example, in 1885, Lyman and others discovered that the

absorption or the emission lines of the hydrogen atom obeyed a simple

formula

where v is the wavenumber of the light, R is the Rydberg constant

(109,677.581 cm-1), and n2and n1 are positive integers associated with the

excited and the ground states, respectively, of the electron in the hydrogen

atom [73, 327] (see Figure 7.9) This formula remained a mystery until

1913, when Niels Bohr proposed a theoretical model of the hydrogen atom

based on the experimental data obtained by Rutherford and the theoretical

concept of the quantum of action invoked by M Planck in 1900 Bohr’s

atomic model led to the correct interpretations of the meanings of n2and

n1 in Eq (7.10) and to the calculation of the Rydberg constant from

fun-damental constants of physics

The atomic absorption spectroscopy discussed above suggests an

interesting analogy:

“cDNA array technology may be to the cell biology of the 21st century

what the line spectroscopy was to the atomic physics of the 20th

This and other related comparisons are summarized in Table 7.1 This

table is not meant to be exhaustively complete but lists only those items

related to the theoretical cell biological research that the author has carried

out during the past four decades and thus does not include many important

contributions made by other researchers, for example, the work of Craig

Benham on SIDDs (stress-induced duplex destabilizations) which are

directly related to the concept of conformons [79, 80, 226]

Figure 7.9 Energy levels of the hydrogen atom [73, p 475].

The term “ribonoscopy” appearing in the third row and the third

col-umn is here defined as the experimental technique that allows biologists

to study genome-wide (i.e., over the whole set of genes in a cell) changes

in the levels of the RNA (ribonucleic acid) molecules inside the cell

meas-ured by cDNA arrays (also called microarrays) and other methods as

functions of environmental perturbations So defined, ribonoscopy may be

viewed as the experimental technique for doing “ribonomics”, a term

coined by Keene meaning the genome-wide study of RNA changes in

cells [328] In other words, ribonoscopy may be to ribonomics what

atomic spectroscopy was to atomic electronics.

“Ribons” appearing in the fifth row and the third column is defined as

the genome-wide spatial and temporal patterns of mRNA levels or

Table 7.1 An analogy between atomic physics and cell biology based on the similarity

between line spectroscopy in atomic physics and cDNA microarray technology in cell

biology.

Time 19th–20th century 20th–21st century

Experimental

technique

Atomic absorption/emission spectroscopy (19th century)

cDNA array technology (“ribonoscopy”) (1995) [307–313]

Experimental data Atomic line spectra mRNA levels in the cell

Regularities Lyman series

Balmer series Ritz-Paschen series Brackett series Pfund series

RNA metabolic modules (ribons) (?) Genetic networks (?)

Cell metabolic networks (?)

Theoretical model Bohr’s atom (1913) The Bhopalator (1985) [15, 16]

Basic concepts Quantum of action (1900) The conformon (1972) [6, 14, 65]

IDSs (1985) [25, pp 69–74]

Cell language theory (1977) [19–23]

Theory Quantum theory (1925) The conformon theory of

molecular machines (1974) Cell language theory (1997) Molecular information theory (2004) [273]

Philosophy Complementarity (1915) Complementarism (1993) [24, 50]

A unified theory of

physics, biology,

and philosophy

A theory of everything (e.g., the Tarragonator (2005) [279])

concentrations inside the cell (such as exemplified by the RNA trajectories

shown in Figure 7.4) Since the mRNA levels are determined by both the

TR and TD (see Eq (7.7)), ribons are species of IDSs (see Section 6.1.2)

The advantage and the utility of the term ribons derive from the fact that it

is directly connected to the rich results of the theories of dissipative

struc-tures worked out by Prigogine and others in the 1980s [58, 59].

7.3 Analysis of Human Breast Cancer Microarray Data

The human breast cancer RNA data analyzed below were obtained by

Perou et al [306] from the human breast tissues biopsied from the normal

tissue (N), tumor before (BE) drug treatment (doxorubicin, 16 weeks),

and tumor after (AF) drug treatment in vivo (Figure 7.10) The fourth

sample from tumor after (AF’) drug treatment in vitro was not obtained by

Perou et al but would be needed if the ribonoscopic method described

here is to be utilized for personalized medicine

7.3.1 The Mechanism Circle-Based Analysis

The human breast cancer data measured by Perou et al [306] can be

organized as shown in Table 7.2 The original mRNA data of Perou et al

[306] in the three columns denoted as N (normal), BE (tumor before

treat-ing with the anticancer drug, doxorubicin), and AF (tumor after treattreat-ing

with drug) in Table 7.2 are processed through steps (1)–(3) and the results

are presented in Figure 7.12

Figure 7.10 The four types of tissues that are required to generate the molecular data,

e.g., RNA sequences and differential expression patterns measured with microarrays or

equivalent techniques N, BE, and AF are needed for generating the molecular data (see

Table 7.2) for theragnostics (i.e., therapeutic and diagnostic purposes), while N, BE, AF,

and AF’ will be needed to generate the molecular data for personalized therapy and

per-sonalized medicine or precision medicine For the sake of simplicity, the symbol AF is

used to indicate either AF or AF’, whenever no confusion can arise under the context of

the discussion involved.

Tumor Tissue After

Drug Treatment in vitro

( AF’ )

Patient X

Biopsy

Normal Tissue Tumor Tissue Before Tumor Tissue After

Drug Treatment Drug Treatment in vivo

( N ) ( BE ) ( AF )

Tissue culture

(1) Calculate the angle a defined as

where ∆D is the change in the RNA levels in the tumor tissue after

drug treatment, i.e.,

and ∆T is the change in the RNA levels induced by tumor, i.e.,

(2) Using the mechanism circle (Figure 7.11), convert the angles a into

their corresponding mechanism numbers based on the rules given in

Table 7.3 to generate the “mechanism table” (Table 7.2)

Table 7.2 The “unfiltered mechanism table”.

N = the number of patients; n= the number of ORFs; SM = survival months; imTI = individual

micro-therapeutic index by Eq (7.15); ITI = individual therapeutic index (see below) N = normal, BE

before drug treatment; AF = after drug treatment; M = mechanism define in Table 7.3 and Figure 7.11

The numbers in the interior of the table are arbitrary one selected for an illustrative purpose only.

Notes: ORF = Open reading frame; N = Normal tissue; BE = tumor tissue before treating with

anticancer drug; AF = tumor tissue after treating with anticancer drug; M = mechanism number from

the Mechanisms Circle; imTI, ITI = individual micro-therapeutic indexes, before and after filtering,

respectively.

Figure 7.11 The mechanism circle The angle a is calculated based on Eq (7.12) and the

meanings of the mechanism numbers are given in Table 7.3.

Table 7.3 The definition of the mechanism numbers and their meanings.

-9 Defined as the mean ± 10% of

the range of angles excluding those lying outside of the mean by 2 σ’s

Note: The symbols are defined thus: + = increase; - = decrease; 0 = no change.

(3) From the a values interpreted in terms of the mechanisms defined in

Figure 7.11 and Table 7.3, construct the “unfiltered mechanism table”

(Table 7.2) The individual micro-therapeutic index (imTI) in the

table is defined as

imTI = (4 + 8)/(2 + 6), (7.15)

where the Arabic numeral x represents the number of the open

read-ing frames (ORFs) whose transcripts exhibit mechanism x in a given

patient, i.e., the number of times x appears in the M column in a given

patient in Table 7.2

(4) Plot the imTI values against the SM (survival month) values from

Table 7.2 to obtain the “survival month vs imTI” plot and the

associ-ated repression line (Figure 7.13, upper panel)

(5) Find an objective method (e.g., the PDE-based method shown in

Figure 7.17) to remove those ORFs whose transcripts exhibit

mecha-nisms 2, 6, 4, or 8 in a given patient so that the unfiltered imTI vs

SM plot in Figure 7.13 (upper panel) can be transformed into the

fimTI vs SM plot shown in the lower panel of Figure 7.13, where

fimTI indicates “filtered micro-therapeutic index” Discovering what

is here referred to as “filtering” would constitute one of the major

objectives of ribonomics (i.e., the study of genome-wide RNA levels)

as applied to cancer research One such method, which is based on

utilizing PDE, is described in Sections 7.3.2 and 7.3.3 (see especially

Figures 7.18, 7.20, and 7.22)

Figure 7.13 lists the results of analyzing 30 genes randomly selected

out of 4,740 genes from each of the 20 patients Mechanisms 2 and 6

indicate that both the breast tumor and doxorubicin induce the RNA level

changes that are in the same direction in the mechanisms circle (Figure 7.11)

and hence are likely to be harmful (but not proven) thus being colored red

(symbolizing a potential danger), while Mechanisms 4 and 8 indicate that

both breast tumor and the drug induce the RNA level changes that are in

the opposite directions in the mechanisms circle and hence are likely to be

beneficial (although not proven) thus being colored green (symbolizing a

potential benefit) In the absence of independent evidence, the words

“harmful” and “beneficial” may be better replaced with the terms such as

“parallel” and “anti-parallel” or with “red” and “green” that are

non-committal as to the clinical (in contrast to molecular-theoretical)

signifi-cance of these mechanisms

The mechanisms defined in Table 7.2 represent the phenotypes on

whole cell levels, since the effects of tumor and drug treatment on the

intracellular levels of individual mRNA molecules would be determined

by the metabolic state of the whole cell In contrast, the Survival Month

(SM) data of breast cancer patients after drug treatment would

repre-sent the phenotypes on the whole human body level, since the life and

death of an individual is determined ultimately by the physiology of the

whole human body, although mRNA levels of the breast tissues of

breast cancer patients can contribute significantly to the cause of their

deaths Thus, it may be necessary to distinguish between at least two

types of phenotypes — the phenotype on the whole-cell level and the

phenotype on the whole-body level — the former may be referred to as

the whole-cell phenotype (WCP) and the latter the whole-body

pheno-type (WBP) As will be discussed in Sections 7.3.2 and 3.3.3, the

rela-tion between WCP and WBP appears to be not one-to-one but rather

one-to-many For example, the WCPs, Mechanisms 2 and 6, which are

likely to be beneficial to patients judged from the perspective of cell

metabolism (since these mechanisms implicate mRNA changes that are

in the same direction whether caused by tumor or drug treatment), are

in fact found not to be so when compared against the SMs of breast

cancer patients That is, when their associated mRNA data are analyzed

based on the PDE as described in Sections 7.3.2 and 3.3.3, Mechanism

2 is found more frequently among long survivors than among short

survivors, while Mechanism 6 is found less frequently among the long

survivors than among short survivors, the former being opposite to

what is expected solely based on the mechanism phenotypes or WCP

alone, although the latter turned out as expected on the basis of WCPs

(see Figure 7.20)

The blue curves in Figure 7.14 are the histograms constructed based

on the frequency distributions of the red (Mechanisms 2 and 6) and green

(Mechanisms 4 and 8) boxes in Figure 7.13 As evident, the blue curves

fit the Poisson distribution almost perfectly The Poisson distribution, Eq

(7.16), is a discrete probability distribution that expresses the probability

of a given number of events, k, occurring in a fixed interval of time and/

or space if these events occur with a known average rate, μ, and

independ-ent of the time since the last evindepend-ent [329]

Figure 7.14a and b indicate that the average rate (~6) of observing

Mechanisms 4 and 8 is higher than the average rate (~ 2) of observing

Mechanisms 2 and 6 Since on average doxorubicin has beneficial effects

on breast cancer patients (otherwise doxorubicin would not have been

selected as a drug), we can conclude that Mechanisms 4 and 8 are more

beneficial (or less harmful) than Mechanisms 2 and 6 on average when

doxorubicin was given to patients

Another evidence supporting the hypothesis that Mechanisms 2 and 6

are indeed harmful and Mechanisms 4 and 8 beneficial comes from

analyzing the breast cancer data using PDE as described in Sections 7.3.2

and 7.3.3

Several conclusions can be drawn from the mechanism table in

Figure 7.12:

Figure 7.12 The RNA level data of Perou et al [306] were processed using the

mecha-nisms circle (Figure 7.11) to reveal the therapeutic effects of doxorubicin on 20 breast

cancer patients Green = antiparallel (or likely beneficial); red = parallel (or likely

harm-ful) A randomly selected partial list out of about 5000 genes (or ORFs).

Figure 7.13 The micro-therapeutic index (mTI) vs SM plot Upper = actual; lower =

hypothetical mTI = (4 + 8)/(2 + 6) = (# of Green Boxes)/(# of Red Boxes in Figure 7.12)

(1) There are no genes whose transcripts exhibit the same mechanism

phenotype, either red (potentially harmful) or green (potentially

ben-eficial), for all 20 patients In other words, there are no continuous

red or green horizontal strips that are unbroken in Figure 7.12,

lead-ing to the followlead-ing generalization:

“There may be few (less than ~0.1%?) genes whose transcripts exhibit

the common dissipative structures (or mechanism phenotypes) in all

Since, according to the IDS-cell function identity (ICFI) hypothesis

explained in Section 3.2.1, IDSs such as RNA trajectories (shown in

Figure 7.4) determine cell functions, we can transform Statement (7.17)

into Statements (7.18) and (7.19):

Figure 7.14 The Poisson distributions of antiparallel and parallel mechanisms (or RNA

dissipatons) in 20 human breast cancer patients Data from Perou et al [306] (a) The

Poisson distribution of mechanisms 4 and 8 (b) The Poisson distribution of mechanisms

2 and 6.

ͲϬ͘Ϭϱ Ϭ Ϭ͘Ϭϱ Ϭ͘ϭ Ϭ͘ϭϱ Ϭ͘Ϯ

ͲϬ͘ϭ Ϭ Ϭ͘ϭ Ϭ͘Ϯ Ϭ͘ϯ

(a)

(b)

There may be no genes that are commonly responsible for all breast

There may be no breast cancer genes or breast cancer genotypes (7.19)

If Statement (7.19) can be generalized and extended to other forms of

cancers, we can conclude that

There are no cancer genes or cancer genotypes (7.20)

(2) Although there are no common genotypes responsible for breast

cancer (cf (7.8) above),

There appear to exist common mRNA phenotypes (i.e., mRNA

dissi-patons) that are closely associated with breast cancer (7.21)

One experimental support for Statement (7.21) is provided by the fact

that any one of the four mechanism phenotypes, i.e., 2, 4, 6, and 8, can be

associated with or “realized by” two or more genes either within a given

patient (see columns in Figure 7.12) or in different patients (see rows in

Figure 7.12) The difference between genes and RNA phenotypes (e.g.,

mechanism phenotypes defined in Figure 7.11) may be compared with the

difference between words and their meanings The same meaning can be

conveyed by two or more different words This is equivalent to saying that

two signs can have the same meaning (or interpretant to use the Peircean

idiom; see Section 6.3 and Figure 9.1) The RNA phenotypes (also called

RNA trajectories, RNA expression profiles, RNA dissipative structures, or

RNA dissipatons, mechanism phenotypes) found in individual breast cancer

patients may be referred to as the patient-specific breast cancer-associated

RNA profiles or patient-specific breast cancer-associated RNA dissipatons.

(3) There are no continuous, unbroken vertical strips of either color, red

or green, in Figure 7.12 This observation, when combined with the

ICFI hypothesis described in Section 3.2.1, can lead to the following

If Statement (7.23) can be generalized, Statement (7.24) would result:

The therapeutic efficacy of anticancer drugs depend on individual

If Statement (7.24) can be substantiated by further studies, it would

provide the empirical basis for advocating the necessity for personalized

medicine (see Figure 7.2c), in contrast to “group” or “average” medicine

7.3.2 PDE-Based Method for Identifying Patient-Specific

Breast Cancer Genes

The PDE was derived from the Planck radiation equation (PRE) in 2008

[25, pp 343–68] by replacing the universal constants and temperature

with free parameters, A, B, and C (see Eqs (8.1) and (8.3) in Figure 8.1)

The unusual feature of PDE is that it fits almost all long-tailed histograms

generated in physics, biology, neuroscience, economics, and linguistics

(see Chapter 8), just as the Gaussian distribution equation fits normally

distributed histograms [330]

The procedure or the algorithm for applying PDE to analyzing human

breast cancer data consists of four main steps as summarized in Figure 7.15:

(1) Transform selected portions of the genome-wide mRNA data into

histograms using the histogram software available in Excel The

selection criteria can be (i) random (Figure 7.18), (ii) based on the

mechanism phenotypes defined in Figure 7.11 (Figure 7.19), or (iii)

based on metabolic pathways (Figures 7.16 and 7.21)

Figure 7.15 The procedure (or algorithm) for analyzing DNA microarray data using

PDE áSMñ = the average survival months of breast cancer patients after drug treatment.

Drug-induced ∆slope vs <SM> plot

1

2

3

4

(2) Using the Solver software available in Excel, determine the

numeri-cal values of A, B, and C of PDE, Eq (8.3) (Figure 7.16).

(3) Plot A vs C and determine the linear equation, y = ax + b, and the

associated correlation coefficient R2 values (e.g., Figure 7.18) If the R2

< 0.60, terminate the analysis and otherwise continue to the next step

(4) Plot the average survival month, SM , of each group vs the

drug-induced change in the slope of the A vs C plots in (3) (Figures 7.18,

bottom panel, and 7.20)

As shown in Figure 8.3(e), the genome-wide RNA levels measured in

human breast tissues from 20 patients fitted PDE almost perfectly Rather

than making one histogram out of the genome-wide RNA levels, we

inves-tigated the RNA levels of a few metabolic pathways shown in Table 7.4

Some examples of the fitting of these pathways to PDE are displayed in

Figure 7.16 Compared to the histogram of the whole genome (with ~5000

genes or ORFs), those of the 3 metabolic pathways (each containing

Figure 7.16 The fittings of PDE to the mRNA levels of the human breast cancer tissues

The x-axis represents the RNA level bin numbers and the y-axis represents frequency CGI

= kinase binding protein; MAPK = mitogen-activated protein kinases; ZFP = zinc finger

proteins; WG = whole genomes of 20 patients (92,813 mRNA levels) The typical PDE

parameter values are given in Table 7.5.

džƉĞƌŝŵĞŶƚĂů W

ZFP

WG

50–100 genes) show considerable noise and yet they all fit the PDE

rea-sonably well The shapes of the PDE curves appear qualitatively different

from one another This impression is confirmed quantitatively when we

compared the PDE parameter ratios, b/A, pair-wise among the six groups

as shown in Table 7.5, since the pair-wise p-values are all less than 0.05

This indicates that

PDE is capable of quantifying the qualitative differences between the

shapes of long tailed histograms that are difficult to distinguish visually

(7.25)

Furthermore, PDE is capable of detecting the subtle effects of

doxo-rubicin treatment on the RNA distributions of certain (but not all)

meta-bolic pathways in the human breast tissues as shown in Table 7.6 Of the

Table 7.4 Protein families studied in this section and their cellular functions.

Kinase-binding protein (CGI) Telomere uncapping and elongation

Unknown proteins (KIAA) Function unknown

Mitogen-activated protein kinase (MAPK) Cell proliferation and survival

Zinc finger protein (ZFP) DNA transcription

Electron-transferring flavoprotein (ETF) Fatty acid oxidation

Table 7.5 The p-values for the pair-wise comparisons among the PDE parameter

values of the five metabolic pathways of the human breast cancer tissues AF =

After drug treatment.

p-values (AF, b/A)

CGI — 1.3E -3 0.04 2.62E -4 7.65E -4 0.02

Note: The RNA levels were those measured from tumor tissues after (AF) drug treatment

Only the b/A ratio is examined in this table For other ratios, b/B and B/A, see Table 7.6.

Table 7.6 The PDE parameter value ratios for the five metabolic pathways of the human breast tissues.

Notes: CGI = kinase binding protein; MAPK = mitogen-activated protein kinases; ZFP = zinc finger proteins; CD = cluster of differentiation; ETF = electron

transferring flavoproteins The p-values were calculated using the Student’s t-test in Excel BE = before treating with doxorubicin; AF = after treating with

doxorubicin.

six groups of RNA levels analyzed with PDE, only one pathway, i.e., the

CD pathway, showed statistically significant changes in all the three PDE

parameter values induced by doxorubicin treatment It is interesting to

note that the statistically significant drug effects on the MAPK pathway

are captured in the b/A ratios, while those on the ETF pathway were

cap-tured in the B/A ratios of PDE Thus, it may be concluded that,

Of the 5 metabolic pathways examined, the cluster of differentiation

proteins may be most intimately connected with breast cancer (7.26)

The 20 breast cancer patients in Table 7.2 were divided into three

groups — (i) short survivors (8–17 months), (ii) intermediate survivors

(22–57 months), and (iii) long survivors (66–89 months) The RNA levels

of the short and long survivors before (BE) and after (AF) treating with

doxorubicin were used to generate long-tailed histograms, some examples

of which are being shown in Figure 7.17 The numerical values of the

Figure 7.17 Some examples of human breast cancer microarray histograms fitting PDE

The mechanisms and the number of ORFs graphed are indicated along with the

informa-tion about drug treatment.

-2 0 2 4 6 8 10 12 14

-10 0 10 20 30 40 50 60 70 80

Figure 7.18 PDE-based analysis of a randomly selected 300 gene transcripts (i.e.,

mRNAs) measured from 20 breast cancer patients before (BE) and after (AF) treating with

doxorubicin for 16 weeks Out of more than a dozen similar analyses of 300 gene

tran-scripts, only 15–20% showed correlations with R2 > 0.6 at the level of the survival month

vs drug-induced changes in the A vs C plots.

vors (BE)

= 0.0005x + 3.478 R² = 0.9688

3 2 4 6 8 4

0.000E+00

2.00E+02 4.00E+02

Sho

0 2 4 6

Interm

0 2 4 6 8

Lon

y = -12.484x + 6 R² = 0.929

ϲ

x10 4

er

y = 0.0012 R² = 0

8.00E+02 1.00E+0

rs (AF)

y = 0.001x + 2 R² = 0.982

urvivor (AF

= 0.0005x + 3.16 R² = 0.9466

r (AF)

2x + 2.6578 0.9562 03 1.20E+03

2.7572 22 50E+03

parameters A and C were then plotted as shown in Figure 7.18 As evident

in Figure 7.18, all the A vs C plots show excellent linear correlations with

the R2 values greater than 0.93 The A vs B plots (not shown) did show

similarly excellent correlations The following features are evident in

Figure 7.18:

(1) The linearity of the A vs C plots indicate that the PDE parameters, A

and C, are tightly coupled.

(2) Since A appears in the first term of PDE and C in the second term and

since the first term is most likely related to the number of standing

waves in the system involved and the second term to the average

energy of the standing waves (in analogy to Planckian radiation

equa-tion, Eq (8.1)), it seems reasonable to postulate that the close

cou-pling between the numerical values of A and C indicates a close

coupling between the standing waves (which are thought to be related

to the organization of breast tissue and hence to the function of the

system, see Figure 8.8) and the energy (likely related to energy

metabolism of individual cells) content of the system Since

organi-zation is a form of work, it must dissipate energy, thus justifying the

correlation between the A and C terms.

(3) The slope of the regression lines in the A vs C plots in Figure 7.18

vary from 1.2 × 10−3 to 0.5 × 10−3, which may be related to the

effi-ciency of tissue organization, normal tissues most likely being more

efficient than tumor tissues

(4) When the SM are plotted against the drug-induced changes in the

slope (∆slope) of the A vs C plots, a linear correlation with a negative

slope was found in Figure 7.18 (see the bottom panel) Since the

x-axis encodes the effects of doxorubicin on the SMs of 20 breast

cancer patients, the negative slope indicates that the drug effect (as

mediated by the mRNA phenotypes of the randomly selected 300

genes) is harmful to patients

(5) When similar analysis as in (4) is carried out with different sets of

300 genes randomly selected, about 10–20% of the sets tested

showed excellent linear correlations, some with positive and some

with negative slopes, the positive slope indicating that some genes

have RNA phenotypes that are beneficial to breast cancer patients,

while the negative slope indicating that some genes exhibit RNA

phenotype that are harmful to breast cancer patients

In another series of the PDE-based analysis of the human breast

can-cer mRNA data, the distributions of the mechanism phenotypes, 2, 4, 6,

and 8, were graphed as histograms which were then fitted to PDE,

produc-ing the numerical values of the parameters, A and C In Figure 7.19, the

mRNA levels exhibiting Mechanisms 2, 4, 6, and 8 before and after

treat-ing with doxorubicin in five long survivors’ histograms were transformed

into histograms which were fitted into PDE, thus producing the numerical

Figure 7.19 A vs C plots of the PDE parameters fitting the mRNA histograms measured

from five long survivors.

LJсϬ͘ϬϬϭϭdžнϯ͘Ϯϰϲϳ ZϸсϬ͘ϲϳ

0 5 10

Ϭ ϱ ϭϬ

Ϭ Ϯ ϰ

Ϭ ϱ

A

Mechanism 8 (AF)

values, A, B, and C From these, the A vs C plots were obtained as shown

The correlation coefficient values for the linear regression lines for the A

vs C ranged from 0.28 to 0.98 Similar analyses were carried out with the

mRNA data measured from five intermediate and five long survivors

From the A vs C plots of the three groups, the SM vs the changes in the

slope (∆ slope) graphs were constructed as shown in Figure 7.20 Except

Mechanism 4, all the plots gave excellent correlation coefficients, i.e.,

greater than 0.92 It is interesting to note that Mechanisms 2 and 8 show

positive slopes, while Mechanism 6 gives a negative one, suggesting that

Mechanisms 2 and 8 are beneficial, while Mechanism 6 is harmful to

patients

When the number, n, of the ORFs analyzed in terms of a histogram is

smaller than about 50, the shapes of the histogram is quite noisy (see

Figures 7.16 and 7.17) However, instead of plotting n mRNA levels, if the

differences between all possible pairs between the n values are utilized,

much smoother histograms are obtained that fit to PDE with greater

preci-sion as shown in Figure 7.21 The number of all possible pairs of n

num-bers is n(n - 1)/2, which is almost one-half of n2 When the number of

elements of a histogram is increased from n to n(n - 1)/2, the correlation

coefficient of the A vs C plots of the PDE fitting the histogram improved

from an average of 0.424 to 0.972 in Figure 7.21 The PDE-based analysis

Figure 7.20 The SM vs ∆slope plots of the three groups of breast cancer patients who

survived — short (10.7 months), intermediate (31 months), and long (75 months) periods

after doxorubicin treatment for 16 weeks SM = the average survival month.

y = 81538x + 43.369 R² = 0.9892 Ϭ

ϮϬ ϰϬ ϲϬ ϴϬ

-0.0006 -0.0004 -0.0002 0 0.0002 0.0004 0.0006

∆ SLOPE Mechanism 2

y = -14475x + 32.143 R² = 0.3255

Ϭ ϮϬ ϰϬ ϲϬ ϴϬ

Ϭ ϮϬ ϰϬ ϲϬ ϴϬ

0 0.0005 0.001 0.0015 0.002 0.0025 0.003 0.0035 0.004

∆ SLOPE Mechanism 6

y = 39033x + 9.309 R² = 0.9198 Ϭ

ϮϬ ϰϬ ϲϬ ϴϬ

0 0.0005 0.001 0.0015 0.002

∆ SLOPE Mechanism 8

of mRNA data using the original number n is referred to as the “absolute

method”, while that using the n(n - 1)/2 values as the “difference method”

in Figure 7.21

The difference method of constructing the mRNA histogram was used

in analyzing the 50 ORFs encoding the KIAA pathway and the resulting

A vs C plots are shown in Figure 7.22 When the drug-induced changes

in the slope of the A vs C graphs are plotted against the average SMs of

the short, intermediate, and long surviving groups, a reasonable linear

cor-relation with a negative slope was found (see the bottom panel in Figure

7.22), indicating that the proteins encoded by the KIAA pathway is

harm-ful to breast cancer patients when averaged over five patients

Figure 7.21 The absolute (old; n) vs difference (new; n(n - 1)/2) methods for

construct-ing histograms.

LJсͲϬ͘ϬϮϰϴdžнϱ͘ϰϰϳϲ ZϸсϬ͘ϲϲϬϭ Ϭ

Ϯ ϰ ϲ

2 4 6

0 1 2 3 4

Figure 7.22 The PDE-based analysis of the mRNA levels of the 15 breast cancer patients

encoding the KIAA pathway.

y = 0.0027x + 2.4148 R² = 0.9721

0 1 2 3 4

2.7 2.8 2.9 3 3.1 3.2 3.3 3.4

0 0.5 1 1.5 2 2.5 3 3.5 4

0 10 20 30 40 50 60 70 80

-0.0015 -0.001 -0.0005 0 0.0005 0.001 0.0015 0.002 0.0025

Drug-induced changes in the slopes of A vs C plots

KIAA Protein (short, intermediate, and long survivors)

7.3.3 Can PDE Be to Cell Biology What PRE is to Atomic Physics?

In Section 7.3.2, PDE has been shown to provide a quantitative method

for identifying the genes, some of whose transcript phenotypes are

correlated to the average SMs of groups of breast cancer patients

(Figures 7.17, 7.19, and 7.21) As an attempt to understand the possible

significance of this finding, I am inclined to suggest that

PDE is to cell biology what the Planck radiation equation was to

Statement (7.27) is in agreement with the atom–cell isomorphism

postulate discussed in Section 3.2 and [25, pp 279–90]

Three important observations can be derived from Section 7.3.2:

(1) Most, if not all, mRNA histograms generated from human breast

cancer tissues, before and after treating with doxorubicin, fit PDE

(see Figures 7.16 and 7.17)

(2) About 30–50% of all the A vs C plots generated from the PDE-fitting

mRNA histograms produced correlation coefficients greater than 0.6

(see Figures 7.18, 7.19, and 7.21)

(3) Many mRNA histograms produce the A vs C plots whose slopes

change due to drug treatment (see Figures 7.18, 7.19, and 7.21), thus

giving rise to the concept of drug-induced ∆slopes that are often

linearly correlated with SM of different patient groups (see

Figures 7.20 and 7.22)

Observation (1) may simply mean that the mRNA level data measured

from human breast cancer tissues are not random but organized due to

some selection processes during human evolution and development of

individuals (cf Section 8.4.1) In other words, the mRNA data measured

from the human breast cancer patients by Perou et al [306] represent the

“organized complexity” of Weaver [331]

Observation (2) indicates that the physicochemical processes

affect-ing the first and the second terms of PDE, i.e., A and C, respectively, are

often coupled, if not always The possible mechanism underlying this A–C

coupling may be inferred based on the analogy between the PRE and PDE

as summarized in Table 7.7 One of the key points of the table is that PRE

is concerned with electromagnetic waves, whereas PDE, as applied to

biology, is primarily concerned with chemical concentration waves, both

obeying the Fourier theorem [53, 160] Based on this assumption, it seems

reasonable to speculate that the first term of PDE is related to the

concen-tration waves of diffusible molecules and ions in the tumor tissues that are

affected by their anatomical organizations including extracellular

matri-ces, blood vessels, lymphatics, and nerve fibers, and the second term is

concerned with the energy metabolism in individual cells (see the last row

in Table 7.7) The former may be related to the Tissue Organization Field

Theory (TOFT) of cancer formation [332] and the latter to the Somatic

Mutation Theory (SMT) [333] TOFT asserts that cancer arises from

dis-organized tissues consisting of millions of cells, while SMT maintains

that cancers originate in mutated genes in individual cells If the

interpre-tation of PDE suggested in Table 7.7 is correct, both these theories may

not be mutually exclusive but are implicated in tumor formation and

maintenance, and the extent of the involvement of these competing

theo-ries of carcinogenesis may be explored using PDE

Observation (3) suggests that PDE-based analysis can reveal those

genes whose RNA phenotypes are implicated in drug-induced effects

(encoded on the x-axis) on the longevity (encoded on the y-axis) of breast

Table 7.7 A qualitative interpretation of PDE in analogy to PRE.

Quantum Mechanics Biology Long-tailed histograms

fitted by (discovered in)

PRE (1900) PDE (2008)

Mathematics U( λ,T) = (2hc/λ5 )/(ehc/ λkT-1) y = (A/(x+B)5 )/(eC/(X+ B)-1)

Waves [362] Electromagnetic waves Chemical concentration waves

Interpretation First term = the number of

standing waves [122]

First term = the number of

standing chemical centration waves (tissue organization field theory [332]) Second term = the

con-average energy of the standing waves [122]

Second term = the energy

metabolism in individual

cells (SMT [333])

cancer patients (Figures 7.20 and 7.22), the positive slope of the graphs

indicating beneficial effects of the drug-induced RNA phenotypes and the

negative slope indicating the harmful effects

7.3.4 The PDE-Based Approach to Discovering Dissipative Structure

(or Dissipaton)- Targeting Drugs

The basic premise underlying the PDE-based approach to drug discovery

that I have been advocating since 2012 [25, p 618] and described below

is that dissipative structures (or dissipatons) (Section 2.6) are the ultimate

targets of drugs in contrast to the traditional view which regards

targeting drugs (DTDs) can be expressed in several equivalent ways:

The ultimate targets of all drugs are the dissipative structures of the

living cell or ic-dissipatons (7.28)

where ic stands for “intracellular”

No therapeutic nor toxic effects can be exerted by any agent without

affecting cell functions or ic-dissipatons (7.29)

It is impossible for an agent to be therapeutically effective unless it can

affect cell functions, i.e., ic-dissipatons (7.30)

Statements (7.28)–(7.30) were referred to as the First Law of

Therag-nostics in 2012 [25, p 618]

Figure 7.23 summarizes the key steps involved in the PDE-based

analy-sis of ribonoscopic data for drug discovery research The method of

prepar-ing the samples from which RNA levels are measured usprepar-ing microarrays is

described in Figure 7.10 During drug discovery phase, the samples N, BE,

and AF are required For personalized medicine, AF samples must be

replaced with AF’ samples to identify the most efficacious drugs from a set

of available candidate drugs for individual patients [25, pp 607–20] The

difference between AF and AF’ is that the former represents the mRNA data

measured from tumor tissues of a group (numbering 10–50?) of test cancer

patients treated with experimental drugs, whereas AF’ represents the