Ants, Bees, Genomes & Evolution @ Queen Mary University London

Avoid having to retract your genomics analysis.

Please cite the Winnower version of this article

2016 Edit: slides are from a 15 minute talk representative of this blog at Popgroup december 2015 in Edinburgh:

Biology is a data-science

The dramatic plunge in DNA sequencing costs means that a single MSc or PhD student can now generate data that would have cost $15,000,000 only ten years ago. We are thus leaping from lab-notebook-scale science to research that requires extensive programming, statistics and high performance computing.

This is exciting & empowering – in particular for small teams working on emerging model organisms that lacked genomic resources. But with great powers come great responsibilities… and risks of doing things wrong. These risks are far greater for genome biologists than, say physicists or astronomers who have strong traditions of working with large datasets. In particular:

Invisible mistakes can be costly

Crucially, data analysis problems can be invisible: the analysis runs, the results seem biologically meaningful, and are wonderfully interpretable but they may in fact be completely wrong.

Geoffrey Chang’s story is an emblematic example. By the mid-2000s this young superstar professor crystallographer had won prestigious awards and published high-profile papers providing 3D-structures of important proteins. For example:

But in 2006, others independently obtained the 3D structure of an ortholog to one of those proteins. Surprisingly, the orthologous structure was essentially a mirror-image of Geoffrey Chang’s result.

Rigorously double-checking his scripts, Geoffrey Chang then realized that: “an in-house data reduction program introduced a change in sign [..,]”.

In other words, a simple +/- error led to plausible and highly publishable but dramatically flawed results. He retracted all five papers.

Devastating for him, for his career, for the people working with him, for the hundreds of scientists who based follow-up analyses and experiments on the flawed 3D structures, and for the taxpayers or foundations funding the research. A small but costly mistake.

Approaches to limit the risks

A +/- sign mistake seems like it should be easily detectable. But how do you ensure that experiments requiring complicated data generation and complex analysis pipelines with interdependencies and sophisticated data structures yield correct results?

We can take inspiration from software developers in internet startups: similarly to academic researchers, they form small teams of qualified people to do great things with new technologies. Their approaches for making software robust can help us make our research robust.

An important premise is that humans make mistakes. Thus (almost) all analysis code includes mistakes (at least initially; this includes unix commands, R, perl/python/ruby/node scripts, et cetera). Increasing robustness of our analyses thus requires becoming better at detecting mistakes – but also ensuring that we make fewer mistakes in the first place. Many approaches exist for this. For example:

This list overlaps at least in part with what has been written elsewhere and my coursework material. In my lab we do our best to follow best practices for the bioinformatics tools we develop and our research on social evolution.

Additionally, the essentials of experimental design are long established: ensuring sufficient power, avoiding confounding factors & pseudoreplication (see above & elsewhere), and using appropriate statistics. Particular caution should be used with new technologies as they include sources of bias that may not be immediately obvious (e.g. Illumina lane, extraction order…).

There is hope

There is no way around it: analysing large datasets is hard.

When genomics projects involved tens of millions of $, much of this went to teams of dedicated data scientists, statisticians and bioinformaticians who could ensure data quality and analysis rigor. As sequencing has gotten cheaper the challenges and costs have shifted even further towards data analysis. For large scale human resequencing projects this is well understood. Despite the challenges being even greater for organisms with only few genomic resources, surprisingly many PIs, researchers and funders focusing on such organisms suppose that individual researchers with little formal training will be able to perform all necessary analysis. This is worrying and suggests that important stakeholders who still have limited experience of large datasets underestimate how easily mistakes with major negative consequences occur and go undetected. We may have to see additional publication retractions for awareness of the risks to fully take hold.

Thankfully, multiple initiatives are improving visibility of the data challenges we face (e.g., 1, 2, 3, 4, 5, 6). Such visibility of the risks – and of how easy it is to implement practices that will improve research robustness – needs to grow among funders, researchers, PIs, journal editors and reviewers. This will ultimately bring more people to do better, more trustworthy science that will never need to be retracted.


This post came together thanks to the SSI Collaborations workshop, Bosco K Ho’s post on Geoffrey Chang, discussions in my lab and through interactions with colleagues at the social insect genomics conference and the NESCent Genome Curation group. YW is funded by the Biotechnology and Biological Sciences Research Council [BB/K004204/1], the Natural Environment Research Council [NE/L00626X/1, EOS Cloud] and is a fellow of the Software Sustainablity Institute.

Please cite The Winnower version of this article

June 2, 2015