Dealing with Randomness in Research

Will the new antibody be specific in a Western blot? How easy will it be to crystallize a protein? Will the RNAi knockdown have off-target effects? Will the overexpressed protein be toxic to the cell? How clean will the purification be?

Whether or not an experiment succeeds, how long it takes, and what the results will be are outcomes that are heavily influenced by unpredictable variables. On a grander scale, the entire direction of a project is significantly influenced by variables that cannot be explained, let alone be predicted in advance. One should in fact anticipate that a large component of one’s research is nothing but a gamble. This can lead to frustration and delays in advancing the project. Yet despite the uncertainty, progress is clearly possible and the outcome of a project can be directed into a place of fruition by making the right decisions. The strategy to strive for may not be to eliminate the unexpected, or encounter it with frustration, but to accept it and learn to work with it. Here are some ideas for dealing with the stochastic nature of research:

  • Be attentive: Extract as much useful information as possible from a result and don’t dismiss undesired findings. Knowledge about the experimental system will increase control over future experiments.
  • Anticipate it and be mentally prepared for it. Remember that it is normal for an experiment to give an unexpected result.
  • Re-evaluate the global strategy regularly based on the new state of the project. That way one can impose the direction even if individual outcomes cannot be predicted (see figure).

  • Be willing to adjust course: Acknowledge the Planning Fallacy and work against it. In “Thinking, Fast and Slow”, Daniel Kahneman describes how insiders are often dangerously optimistic about the future of their projects and therefore reluctant to change course. As scientists, we have to be careful about this human weakness and learn to enact change even if it feels painful and uncomfortable.
  • Technical performance: Plan and execute every experiment carefully to minimize stochastic effects due to poor technical performance.
  • Acknowledge delays due to stochasticity: Give yourself enough time while devising the timeline to make room for the delays that are due to unexpected findings.
  • Improve skill and intuition by paying attention while doing experiments. An acute intuition can help navigate through noise and randomness.
  • Plan using all the available knowledge (broad framing), rather than only what you remember about the most recent results (narrow framing). This establishes a global perspective and brings clarity to the planning strategy.

Can Biology Have Its Higgs?

 
Image credit: jager / 123RF Stock Photo

Two weeks ago a spectacular discovery was announced: the Higgs boson, a subatomic particle that endows all matter with mass, was at last discovered. What I found most incredible about this finding was that it confirmed predictions made almost fifty years ago. The discovery was therefore more a validation of a profound postulate than an unexpected new finding. Similar predictions that were followed by experimental validations have been made before – the Periodic Table, first published in 1869, predicted the existence of new atomic elements, many of which were not discovered until late in the 20th century. Einstein’s Theory of General Relativity, published in 1916, predicted the bending of light by mass, to be observed three years later during the 1919 solar eclipse.

But have similar predictions been postulated in Biology, to be validated years later? I would argue that Biology lacks the general principles and firm foundations that are necessary for postulating fundamental principles. Physics can be predictive because the processes it studies can usually be described with mathematical accuracy: Newtonian Mechanics and Maxwell’s Equations, for example, provide us with equations that are highly predictive for the motions of bodies and the dynamics of electromagnetic waves. Even Quantum Mechanics, while non-deterministic, provides statistical equations that describe and predict probabilities.

Image credit: gunnar3000 / 123RF Stock Photo

Biology, on the other hand, is rather messy. In some ways it resembles a language – it evolves in a stochastic fashion with no direction whatsoever. In doing so it doesn’t care about simplicity, but only about preserving the individual and its offspring in the face of natural selection. Cells, organisms and ecosystems acquire complexity and accumulate scores of exceptions to the supposed general rules. These exceptions (for example ribozymes, the platypus or vestigial organs) can be explained but could hardly have been predicted. Physics can afford to be deductive – specific events can be predicted from grand principles-, whereas Biology is mostly inductive – specific events are collected and used to formulate grand principles. Of course, the grand principles derived in Biology are really only general guidelines, whereas the grand principles from Physics are often absolute and free from exceptions.

Even the simplest systems in Biology – the structure of a small protein or the pathway for metabolizing a nutrient – cannot be predicted from first principles, although advances are being made through the use of computer superclusters that can optimize all parameters to yield approximate predictions.

Do you know of postulates about general principles that have been made in Biology, to later be validated  experimentally? What other differences between the predictive capacities of Biology and Physics do you know of?

Competition in Science: To Love it or to Hate it

Why do scientists so often shun competition instead of embracing it as a natural component of their job? This attitute indeed appears to be relatively rare relative to other professions in which competition is accepted and incorporated openly into the working strategy: businesses must beat other businesses for customers, lawyers must compete against other lawyers for clients and for victory in court, politicians either win or loose elections against other candidates, and professional athletes make a living out of defeating other athletes.

Competition arises naturally when a resource is scarce and cannot be provided to all those who desire it. In business, it’s the customers’ purchase and loyalty, in politics the vote, and in sports the medal or the cup. Similarly, the scientific system provides limited access to grants and fellowships, discoveries and publications, jobs at all stages, and qualified students and postdocs for a lab seeking to hire. In some cases, such as when competing for grant applications or when seeking to recruit the best researchers, the identity of the competitors is unknown and little can be done to win except to perform as well as possible. In contrast, when the limited resource is a discovery and its publication, the identity of the competitor is often known and must be taken into account to ensure victory. This would entail not sharing critical data prematurely at conferences and keeping a close eye on what the competition is publishing, similarly to how professional athletes often study the moves of their opponents in great detail.

Is there a reason to think that competition is justified in other professions but not in science? In some instances, competition can be beneficial to scientific progress, by forcing higher quality due to increased scrutiny by the community, and encouraging fast progress for fear of loosing credit for a discovery. The danger is that pressure to publish fast could also make the researcher more sloppy and willing to report results before they have been confirmed by sufficient evidence. Some scientists will go further and resort to dirty methods to defeat their competitors, a practice that has been the subject of many urban legends and is clearly wrong for many reasons (to be discussed in a future post). A danger of excessive competition is also that it can limit the sharing and discussing of ideas and thereby stiffle scientific progress.

So why do scientists so often shun competition? One explanation may be found in the early years during which aspiring scientists in high school or in university first decide to pursue scientific research as a career. The criteria by which the best individuals are accepted into universities and Ph.D. programs are mostly noncompetitive: getting good grades or making a meaningful contribution in summer research projects. And what draws most young scientists into research is not winning or “being the best”, but the sheer joy of learning, making discoveries, and working in a lab. By the time young scientists first confront competition, when they apply for a research scholarship or have begun their Ph.D. in a crowded research field, they are caught by surprise and often react defensively.

Lastly, we shall not forget that competition is not new to science – James Watson has claimed that his and Francis Crick’s discovery of the structure of DNA in 1953 was the product of fierce competition against Linus Pauling, while Marshall Nirenberg and his team at the National Institutes of Health fought hard to defeat Severo Ochoa in the quest for deciphering the genetic code.

The question thus is not whether or not scientists should be competitive, but rather how to engage in healthy competition that leads to more progress and better work without abolishing communication.

Image credit: georgemuresan / 123RF Stock Photo