Article 2: Andrea Califano

By Rita Charon

July 21, 2021
Andrea Califano

A discussion with Andrea Califano and narrative medicine writer Rita Charon, "Aflight in ideas and visions, he and his colleagues in systems biology have been soaring toward what Califano predicts as the Golden Age of Biology. Not unlike the golden ages of physics and chemistry, when Newton and Lavoisier broke through conceptual barriers to see with new minds, this age equips biology for flight"

 

 

Narratives of Discovery

 All non-referenced quotes are the words of Dr. Califano

Art and science unfold in time as chronological sightings overturn until-then received knowledge—from the epic to the novel, from classical portraiture to cubism, from before Copernicus to after Copernicus. Art and science unfold in space as discoverers choose a scale and a scope—from parole to langue, from harmony to symphony, from gene to species, from nano to meta.

Andrea Califano does it all at once. Aflight in ideas and visions, he and his colleagues in systems biology have been soaring toward what Califano predicts as the Golden Age of Biology. Not unlike the golden ages of physics and chemistry, when Newton and Lavoisier broke through conceptual barriers to see with new minds, this age equips biology for flight:

        “The golden age of many sciences—from physics, to economics, to meteorology—has occurred when it finally became possible to transition from an empirical to an actual analytical base. . . . [T]he golden age of biology will likely ensue from our increasing ability to supplement what we can do empirically in the lab with models that can actually predict the results of an experiment without having to perform it.  What became clear in the early 2000s is that, in order to make the next step forward, biology had to transform into a predictive science.”

Founding Chair of Columbia’s Department of Systems Biology, co-founder of DarwinHealth, Director of JP Sulzberger Columbia Genome Center, Co-Leader of the Precision Oncology and Systems Biology program at the Herbert Irving Comprehensive Cancer Center, and discoverer of field-defining break-throughs in cellular mechanisms that govern the cell’s transcriptional states, Califano has helped to fundamentally shift not just cancer biology but biology tout court toward ever more accurate and predictive maps of the inter-workings of the genetics, the epigenetics, the proteomics, and the outcomes of how cells work and even how to alter their workings to the benefit of our health.

As the child of a physical chemist in a family of Italian physicians, Andrea, age 10, built his own chemistry lab in the garage of his Florence house, pilfering glassware and reagents from his father’s laboratory at the university. Drawn to the beauty of organic chemistry while devouring quantum mechanics textbooks as a fourteen-year-old, he troped away from what he calls “trial-and-error” science toward the search for fundamental principles about biological processes. He began his studies in physics at University of Florence and then accepted MIT’s offer of a post-doctoral position in computational mechanics. From there, he chose IBM’s offer of a position at the intersection of computer science, physics, and artificial intelligence, becoming a scientist in the field of complex pattern recognition and machine vision.

An algorithm developed in lab called ARACNe uses gene expression data to identify transcriptional interactions
An algorithm developed in Dr. Califano's lab called ARACNe uses gene expression data to identify transcriptional interactions

Realizing that his visual pattern-recognition algorithms were directly applicable to the study of genomic data, he expanded his own by then authoritative expertise in computer vision to encompass DNA pattern-matching, developing such algorithms as FLASH and SPLASH and CASTOR to recognize genetic patterns and, ultimately, to discover the patterns underlying biological phenomena. By 2000, he was co-authoring many cancer-related papers with Columbia’s medical center faculty and basic science papers on the G-protein-coupled receptors that support odor perception with Stuart Firestein, the Columbia biologist who explained in his 2012 book Ignorance: How It Drives Science that “knowledgeable ignorance, perceptive ignorance, insightful ignorance . . . leads us to frame better questions, the first step to getting better answers. It is the most important resource we scientists have, and using it correctly is the most important thing a scientist does.” [1]

In 2003, Columbia recruited Andrea away from First Genetic Trust, a pharmacogenomic company he had co-founded after leaving IBM in 2000, with the lure of his own systems biology center, in close collaboration with genomic medicine and biomedical informatics.  Tutored in B-cell biology by Director of Columbia’s Institute for Cancer Genetics Riccardo Dalla-Favera, Andrea became a shape-shifter and theorist in science, visionary in his simultaneous fluency in computational biology, physics, and now cancer biology. He used his “insightful ignorance” to study what he calls “the underlying reality of complex biological mechanisms”:

        “If the rules that you set are based on an underlying reality then you can actually start simulating things that are fairly complex and then are predictive of what you would get in an experiment.”

Put another way, by an earlier philosopher of science, Karl Popper in his magisterial Conjectures and Refutation: “This may be expressed by saying—as Schlick did, following Wittgenstein—that a universal law or a theory is not a proper statement but rather “a rule, or a set of instructions, for the derivation of singular statements from other singular statements.”[2]

Andrea was gripped by the mystery of the cancer cell. He told me the story of his  approach:

        “We started by reflecting on a surprising paradox. If you look at the transcriptional state of [tumor] cells, it  is actually very stable, in fact, just as stable as normal cells. If you have a liver cell, you can change temperature and nutrients and even introduce many mutations in its DNA, it will continue to work as a liver cell. It doesn't turn into a B-cell. At the other end of the spectrum, even within the same tumor type, mutations [are] all over the place, so you can get two breast cancers or two prostate cancers with mutations that have nothing to do with each other . . .  So that's a big paradox: how is it possible that mutations are so heterogenous but the cancer state they induce is so homogenous?  In normal cell physiology the reason why cells are able to maintain the stability of their state is their remarkable homeostatic control. So we thought that in cancer there must be an equivalent piece of homeostatic control machinery responsible for integrating the effects of widely diverse mutational repertoires to implement a virtually identical, highly stable cell state.” 

This was the genesis of Andrea’s hunt for Master Regulators, the “piece of machinery” that he fantasized as necessary to explain what he and his colleagues had observed. By now, not only the idea of Master Regulators but the identities and modular structure that allow them to coherently control the transcriptional state of individual tumor cells have become accepted within not only systems biology but in cancer biology and beyond.

        “Studying things one gene at a time,” Andrea told me, “was not going to work. It’s like taking a watch apart and trying to figure out how it works by looking at each individual gear in isolation, without an assembly manual.” To provide a different route to understanding what makes a cell a cancer cell,  Andrea and his colleagues launched predictive biology with the development of such algorithms as ARACNe, CASTOR, and VIPER, the names themselves connoting powerful  biological or mythological heroes embarked on knowledge-finding quests. Utilizing hundreds or thousands of data points to study actual cells, the algorithms can winnow down clouds of data to propose a small number of likely candidates that might satisfy the quest for one’s holy grail, be it the most likely proteins triggering a damaging process in vivo, as in their work on ALS,[3] or the sensitivity of a specific patient’s cancer to a clinically relevant drug.

The architecture of the Master Regulator theory is now well worked out. In a “Perspectives” essay in Nature Reviews Cancer, Califano and Alvarez marshal the evidence to propose that:

cancer cell bottleneck
Schematic representation of how cancer bottlenecks channel upstream genetic alterations to initiate cancer driving programs.

        “Tumour checkpoints represent the cellular logic that is responsible for integrating the effect of large and heterogeneous genomic alteration repertoires into virtually indistinguishable tumour states. This also supports the role of tumour checkpoint (MRs) [Master Regulators] as highly conserved, mechanistic therapeutic targets and biomarkers. Indeed, based on the proposed “Oncotecture” model, tumour checkpoints emerge as the regulatory pillars of tumour dystasis and thus, potentially, as the universal Achilles heels of cancer.”[4]

Andrea’s capacity to see into the future—of biology, of cancer, and of the increasing number of non-cancer diseases like ALS, dementia, diabetes, and alcohol addiction that he and his colleagues have investigated—has already changed the landscape of medical science. His computational “prophesies” have reconfigured not only the ways that researchers investigate particular diseases but are increasingly influencing clinicians’ prescribing practices by predicting algorithmically which master regulators might be responsible for pathophysiological aspects of a particular tumor’s virulence.[5]

This is the shocking gift: It is not just the algorithms and the hundreds of thousands of lines of software that he himself has written. It is rather the clear-eyed resolution of what may well be light-years ahead in biology:

        “A lot of ideas end up being very, very simple-minded in hindsight. [But] it’s very difficult to foresee whether simple-minded ideas will pan out or not in the end. There’s nothing magical about what we do. It’s just that nobody had kind of thought of cancer in this way. . . . And this is not just in cancer. Because this is not only how cancer cells work but rather how virtually all human cells work.”

References

[1] Firestein S. Ignorance: How It Drives Science. New York: Oxford University Press, 2012: 6.

[2] Popper K. Conjectures and Refutations: The Growth of Scientific Knowledge. New York: Basic Books:  108.

[3] Mishra V, Re DB, Le Verche V, Alvarez MJ, Vasciaveo A, Jacquier A, Doulias P-T, Greco TM, Nizzardo M, Papadimitriou D, Nagata T, Rinchetti P, Perez-Torres  EJ, Politi KA, Ikiz B, Clare K, Than ME, Corti S, Ischiropoulos H, Lotti F, Califano A & Przedborski S. Systemic elucidation of neuron-astrocyte interaction in models of amyotrophic lateral sclerosis using multi-modal integrated bioinformatics workflow.  Nature Communications 2020; 11:5579.  https://doi.org/10.1038/s41467-020-19177-y

[4] Califano A, Alvarez MJ. The recurrent architecture of tumour initiation, progression and drug sensitivity. Nature Reviews Cancer 2017;17: 116-130 (p. 126).

[5] Paull EO, Aytes A, Jones SJ, Subramaniam PS, Giorgi FM, Douglass EF, Tagore S, Chu B, Vasciaveo A, Zheng S, Verhaak R, Abate-Shen C, Alvarez MJ, and Califano A. A modular master regulator landscape controls cancer transcriptional identity. Cell 2021;184:334-351.