The Tricky Nature of Research Results

You prepare a sample with interesting properties. Treat it according to a protocol. Measure in some instrument. Then do some calculations on the measured values.

Handled individually, experimental results are conceptually simple (even though the process outlined above may itself be very complex). But when you have a research organisation with lots of scientists and projects generating huge amounts of results (that you also want to combine with external/public data), the challenges of research data management become immense because of the extreme diversity and complexity of the results:

  • There’s variation in the basic types of results: Numbers, colors, hit-lists, images, video-sequences, spectrums, peak-lists, chromatograms, and many more.
  • There’s variation in experimental conditions.
  • There’s variation in result processing and interpretation.
  • There’s variation in selection of algorithms, some results may be excluded, and there may be multiple repetitions of (in principle) identical experiments.

One useful distinction is between “raw data” and “final results” – but this is not a clear distinction, because there may be many steps going from one to the other, with “intermediate results” as partial products, and a given result may be a final result in one context and an intermediate result in another context; e.g. a concentration measurement may be a final result in its own right but also an intermediate result in a batch approval. Registration and visualization of this result hierarchy is an important requirement.

Sometimes you want to treat different types of results in roughly the same way, simply providing a list of all results for a given sample – at other times, you want all the details of each result and the ability to open specific type of software to analyze the raw data behind a result.

When querying and analyzing results in aggregate, it will often be relevant to look at all results for a given compound or all results within a single project, but at other times it’s more relevant to look at all results from a specific assay, or results from different types of analyses run on the same instrument, or results against a specific batch of a reference material.

It may also be the case that the same data have been analyzed in several different ways, and you need to make sure that you are making a valid comparison.

IT Approach

All this complexity in the nature of results invites a heavy handed IT system that requires cumbersome data registration and restrict data processing options in order to accommodate all scenarios – but this approach goes squarely against the needs of scientists for an agile and flexible system that can keep up with quickly changing research workflows. After all, the purpose of the IT system is not to burden but to support research activities and innovation.

Next on the Sciformatics Blog:
Assay Data Management

Stay tuned for next weeks post on a balanced and agile approach to assay data management, enabling high quality data registration without standing in the way of innovative science and research.