Reductionism, Complexity, and Molecular Medicine:
Genetic Chips and the “Globalization” of the Genome
Kenneth F. Schaffner, M.D., Ph.D.
George Washington University
This paper has two parts. Part 1 is somewhat historical but includes material from the history of philosophy of science; Part 2 deals more with current scientific developments but within a philosophical perspective. The historical-philosophical part begins with a brief account of reductionism that sees molecular biology as moving from a period of simplicity and linearity to one involving complexity and global approaches. I then turn to the philosophy of science literature for a more personal view of the relationship between reductionism and genetics in the 1960s, 70s, 80s and 90s, touching on the emergence of the antireductionist consensus in philosophy of biology, and the subsequent reaction that defended a reductionist anticonsensus in biology. Some comments follow describing a
reappreciation of complexity by geneticists in the 1990s, including references to Lander and Schork‟s fourfold way. Here I also discuss examples from cystic fibrosis and cancer genetics, and the shift from what Plomin has termed a “one gene--one disorder” (OGOD)
view to a more recent many-genes--variable range of disorders (MGVRD) perspective.
The second part of this paper considers genetic chips or microarrays as one recent tool developed to deal with the reappreciated complexity of biological organisms, including their genetic complexity. Gene chip technology also resonates with many
geneticists‟ perceived need to move beyond a focus on the DNA sequence and
understand genes in simultaneous complex interactions -- what are called “global views.”
I sketch two examples in which gene chips have been used to implement this global approach, including analysis of a metabolic shift in yeasts and the metamorphosis of flies, and touch briefly on the classification and treatment of human cancers. Finally I consider some lessons, limitations, and prospects for further work in a variety of directions including nascent developments in proteomics.
Part I: Some Historical and Philosophical Developments Relevant to Reductionism--1960-2000
Some themes in molecular biology
One can look at the past fifty years of molecular biology from many different perspectives, but the one that I think is most germane to this conference‟s themes sees biology moving from a period emphasizing simplicity and linearity to one that has become more focused on complexity and global approaches. The Watson-Crick structure for DNA discovered in 1953  was a simple and essentially linear representation of
genomic information, and the genetic code deciphered in 1960 was also linear (albeit redundant) . Early reductionistic strategies such as Benzer‟s 1956 identification of the
“corrected” classical gene with segments of a Watson-Crick DNA sequence were
essentially a one-many identification (more specifically one [Mendelian] gene; three
sequence lengths [muton, recon, cistron]) [3,4]. The process of gene-directed protein
synthesis, encapsulated in the “central dogma” of molecular biology (gene; RNA ;
protein), was similarly simple and linear , and the genetic code was thought to be
universal. Even the Jacob-Monod operon model of 1959-1961, though more reticulate and involving feedback and control genes, underscored the prospect that the diverse complexity found in biology would be explicated using simple on-off interactions at the molecular level.
Developments in the mid-1970s began to cast doubt on simplicity at a molecular level, as recombinant DNA studies and the emerging diversity of protein sequences revealed a rapidly growing multiplicity of molecules and pathways. Successive editions of Watson‟s well-known textbook, Molecular Biology of the Gene, first appearing in
1965, then in 1970, 1976 and 1987 are a microcosm reflecting this increasing developing complexity.
Though molecular genetics was rapidly increasing in complexity at the time, the Human Genome Project (HGP) that was born in the late 1980s represented the apotheosis of a view of simplicity and linearity of genetics in the project‟s early forms . Obtaining
the sequence of DNA for humans was to be a “Holy Grail” that would enable the diagnosis and cure of human disease . But ironically, just as James Watson, the first
director of the National Institutes of Health component of the HGP, was implementing this huge scientific project in the early 1990s, additional tools for interpreting the genetic
code were heralding the need for considerably more complex analyses, tools typified by Lander and Schork‟s “fourfold way,”  a topic to which I shall return to again shortly.
Some Themes Concerning Reduction in Philosophy of Science and Biology
The General Reduction Model and the Antireductionist Consensus. During the early
period of modern molecular biology, essentially its first 25 years from 1953 through 1978, a thesis that molecular biology would provide simple, general, and unified explanations of biology was eminently defensible. One approach to reduction that generalized the classical Nagel model of theory reduction  and incorporated insights of Popper,
Feyerabend, and Kuhn was introduced and defended by the present author in several papers [4, 10, 11, 12] against the criticisms of Hull  and Wimsatt ; for a summary
and extensions see . In a prescient way in his 1974 book , Hull identified some of the complications standing in the way of any simple mapping between Mendelian (transmission) genetics, taken to be typical of biology, and molecular genetics, seen as paradigmatic of a physicochemical reducing science. He argued that the relations between Mendelian and molecular genetics were not one-many, as Benzer had suggested in his modification of the classical concept of the gene in the light of the Watson-Crick model of DNA, but much more bizarre and reticulate -- a “many-many” relationship that
would defeat any simple mappings between those sciences. Hull‟s important criticisms were accepted and elaborated by Kitcher  and Rosenberg , and with the backdrop
of the biology of the 1980s providing prima facie supporting details for an emerging complexity, this view of the non-reducibility of Mendelian to molecular biology had achieved the status of an “antireductionist consensus” by the 1990s [18, 19].
The antireductionist consensus has been concisely and evenhandedly discussed recently by Sterelny and Griffiths , and the reader is encouraged to examine their
account of the back-and-forth arguments that were involved in the discussions (also see Waters‟ reductionist defense  and Sarkar‟s analysis in his ). From my own
perspective, the most salient issues involved in the antireductionist consensus involved three intertwined theses: (1) Replacement of Mendelian genetics, (2) Autonomy of Mendelian genetics, and (3) Mind-boggling many-many complexity interrelating Mendelian and molecular genetics. (“Mind-boggling” is Pat Churchland‟s term for this
kind of view of the complexity of relationships, although she introduces the term in connection with mind-brain relations -- see her .)
In his 1974 book, Hull advanced a curious replacement thesis as part of his
argument that molecular genetics could not reduce Mendelian genetics. Since there were no simple connections between the entities (such as genes) and predicates (like dominant) in the two domains (molecular and Mendelian genetics), a reduction could not be occurring. And if a reduction was not occurring, the only relationship could be that of replacement. This aspect of Hull‟s thesis was intriguing, but was accepted so far as I am aware, by virtually no one (more on this below). But his arguments regarding the difficulties of mapping between entities and predicates did take hold, and were further amplified by Kitcher in his influential 1984 article , and were there used to support a thesis of the partial autonomy of Mendelian genetics. For Kitcher, cell biology, and the process of meiosis in which an account of gene segregation and linkage was given,
provided a reduction of sorts of Mendelian genetics, but no further lower level of analysis was needed or obtainable. Hull‟s and Kitcher‟s accounts were largely accepted by Rosenberg who underscored the difficulties of providing workable connections between Mendelian genes and molecular entities (see both his 1985  and his 1994  books). Many others joined this debate, mostly on the antireductionist side -- for a summary and references see my , esp. ch. 9.
The Reductionist Anticonsensus. Though biology was becoming more complex
throughout the 1980s and 90s, and thus providing encouragement for the philosophical antireductionists‟ thesis of nonconnectability of Mendelian (transmission) genetics and molecular genetics, most molecular biologists themselves perceived of their methods and results as supporting reductionist rather than antireductionist claims, whether this be in genetics or more broadly. In 1987, Eric R. Kandel, a neuroscientist and a recent (2000) Nobel laureate, wrote in the Preface to a volume on Molecular Neurobiology in
Neurology and Psychiatry that:
This volume reflects the impact of molecular biology on neural science and
particularly on neurology and psychiatry. These new approaches have accelerated
the growth of neurobiology. The resulting increase in knowledge has brought with
it two unanticipated consequences that have changed the ways in which clinical
researchers and practitioners can now view the findings that came from basic
science. The first consequence is a new unity, a greater coherence, in biology as a
whole, as studies move from the level of the cell to that of the molecule....
(The second consequence was that “as science becomes more powerful it becomes more ambitious -- it becomes bolder…. And some biological researchers are so bold as to see
their ultimate interest as the function of the human mind.) 
And, as Sterelny and Griffiths note, molecular biologist Gunther Stent wrote that:
What geneticist could take seriously any explication of “reductionism” which
leads to the conclusion that molecular genetics does not amount to a successful
reduction of classical genetics. 
(This short account of the emergence of the antireductionist consensus and the counter- view of the reductionist anticonsensus – the term is, I believe, due to Waters  –
oversimplifies the issues, but to examine them in the depth needed would take us beyond the scope of this article. A more extensive analysis can be found in my  and more recent reflections on the topic will be provided in a forthcoming book.)
What really went on in biology from 1975-2000 (a personal perspective)? There are
different visions among philosophers and biologists and the following is primarily my own personal perspective on developments from 1975 to 2000. In my view, most biologists saw Mendelian (better Mendelian-Morganian/ transmission) genetics as a limiting case of cellular/subcellular/molecular genetics, but adopted various differing strategies to deal with the relations. Watson and his coauthors of the 1987 edition of Molecular Biology of the Gene  took the approach of discussing Mendelian genetics
in chapter 1, then getting on with the real stuff, molecular biology, with virtually no further mention of Mendel or the approaches of transmission genetics. On the other hand, Alberts et al. in their influential 1995 edition of Molecular Biology of the Cell do not
define the classical gene concept until p. 423, and then they modify it on p. 457 to handle alternative RNA splicing, well after simple and complex molecules and DNA, as well as protein synthesis, are introduced . In a 1998 book which is much less molecularly-
oriented, Lynch and Walsh in their Genetics and Analysis of Quantitative Traits see both
quantitative transmission genetics and molecular genetics as “two levels of organization
that will ultimately have to answer to each other, and it is likely they will soon do so.”( p.7) They also add “An ultimate understanding of the mechanisms responsible for expressed variation in quantitative characters requires information at the molecular level.”(p. 321)
I read the history as represented by these three texts as indicating that no biologist (seriously) bought the replacement thesis, though those who were more
molecularly oriented thought and increasingly wrote largely in molecular terms. I also think that no one accepted the autonomy thesis (though Mayr , Lewontin and Levins
, and others argued for various kinds of emergence notions that asserted biology goes beyond purely molecular accounts). We also see a kind of quasi-autonomy thesis in Lynch and Walsh‟s views above. However, increasingly everyone embraced a complexity
(or many-many) thesis, even within molecular genetics, as additional mechanisms and
families of mechanisms and pathways rapidly emerged from a triumphantly advancing science with its new and more powerful instruments for deciphering sequences and studying structural interactions. (I say “even within” molecular genetics because so many
varied mechanisms for related processes were discovered, e.g., multiple ways of regulating DNA translation at the molecular level, even in E. coli.)
But having acknowledged this complexity, biologists began to explore better ways to analyze complexity – no boggled minds here. Some moved to explore
“complexity theories” -- a term that has several different senses. Excellent introductions to several of these senses can be found in Goodwin  and in Liebovitch , which
address these perhaps “sexier” methods, including chaos theory and fractals, than I will
focus on here. In this paper, I consider what I view as more mainstream approaches to working with increasingly complex genetic systems. A paradigm essay in genetics that represents this development is Lander and Schork‟s article that appeared in the special
1994 genome issue of Science .
Complexity in Genetics and the Four-Fold Way.
The 1994 article by Lander and Schork was a landmark essay signaling the re-appreciation of genetic complexity. In the abstract for their article they noted that “medical genetics was revolutionized in the 1980s by the application of genetic mapping to locate the genes responsible for simple Mendelian diseases” ( 2037). But cautioned that:
Most diseases and traits, however, do not follow simple inheritance patters.
Geneticists have this begun to take up the even greater challenge of the genetic
dissection of complex traits. Four major approaches have been developed: linkage
analysis, allele-sharing methods, association studies, and polygenic analysis
[including QTLs] of experimental crosses.(  2037).
Lander and Schork added that “this article synthesizes the current state of the genetic dissection of complex traits -- describing the methods, limitations, and recent applications to biological problems.”
The Lander and Schork article is far too long, and complex in its own right, to summarize in this paper. Suffice it to say that the present author did write, with the help of Irving I. Gottesman and Eric Turkheimer, respectively a genetically-oriented psychologist and a behavioral geneticist, a summary that was intended to be accessible to policy makers interested in complex trait genetics. That paper was presented orally at a national meeting of the American Society for Bioethics and Humanities in 1997, and in updated form it is due for publication in a volume being put together by the American Association for the Advancement of Science and the Hastings Center, and is available from the author on request .
For the purposes of the current paper it is sufficient to note that Lander and Schork found they needed to address the reasons that very few genetic markers show
perfect cosegregation with a complex trait. The reasons included incomplete penetrance and phenocopies, genetic heterogeneity, and polygenic inheritance, the confounding effects of a high frequency of disease-causing alleles, as well as other transmission mechanisms (e.g., imprinting and anticipation). Lander and Schork‟s overview of the “four major approaches” cited above was a tour de force in communicating ways that
geneticists could get purchases on complex traits. A review of the genetics literature of