Keeping scrupulous records and taking the right decision when managing outliers is fundamental to good practice in science. However, in a survey of PRBB scientists, over 40% of the 521 people who replied, said they had sometimes or often noticed problems with record keeping in their immediate working environment. This is not unusual – other scientists in other centres around the world report similar levels of this problem. But being the same as others in sloppy practice is no cause for pride.
During the week starting the 25th of January the PRBB Good Scientific Practice Working Group helped PRBB scientists clean up their act when it comes to how they handle scientific data.
Here you can see the presentations used during the three specialist workshops that took place – focused on wetlab data, bioinformatics data and human data -, as well as the Big Quiz questions and how scientists replied to them. Feel free to join in the conversation by commenting below!
There were three special workshops taking place at the PRBB. Below you can see the program and presentations of each seminar.
Tuesday 26th: 1-3.30pm “Keeping the data record straight in the lab”
Wednesday 27th: 12-3pm “In silico data tsunami: will you survive?”
Thursday 28th: 12’30-3pm “Managing data in human research”
THE BIG QUIZ
Do you really know how to manage your data?
The following questions were used to test the researchers’ knowledge of data management as well as to spark a conversation amongst colleagues – what do they think, what would you do? Some had clear answers, others not so much, and more than one of the options could be correct. See the answers recorded here, together with a comment in response to each question – and do comment below!
What should you include in your lab notebook?
According to the PRBB Good Scientific Practice Code “Without exception, all data arising from experiments or research observations must be recorded (…). The records must also include changes, errors and negative, unexpected, or conflicting results, as well as an indication of the person who performed the experiment or made the observation.”
And: “The individual responsible for the research will have a single record accessible to third parties, of the locations of all samples and data-collection instruments (registered notebooks, databases, etc).”
As a general rule, the research record should be as complete and detailed as possible – including sources of data, experimental details and protocols, ideas and discussions – so that a third person would be able to replicate the research. The record should also include references to where the original data or samples are located (in computers, fridges, etc).
However, confidential data should not be included in the standard records (notebooks, etc.). Rather the notebook should contain a reference to the data, but the confidential data itself should be kept in the appropriate systems and with the relevant security mechanisms.
For how long should you store your data?
The length of time data should be stored varies, but generally speaking it should be long enough after publication to enable the research group or a third person to replicate the experiment or go back to check any details. When the person who has done the research leaves the lab, the data should still be kept in the laboratory.
At the PRBB, the Good Scientific Practice Code suggests a minimum of 5 years of storage: “All original primary information and biological and chemical material arising from a research project must be stored for a minimum of 5 years from the date of the first publication of the results, except in those cases in which the law allows shorter storage periods or requires longer periods to be applied. If the centre allows, the primary information and material may remain stored for longer periods, provided their final destination meets the approval of the person responsible for the research.”
Who owns your project database and lab notebook?
The PRBB Code of Good Scientific Practice states that: “All primary documentation (registered data-collection notebooks, databases, etc.) and biological or chemical material obtained in the course of a research project is the property of the centre to which the person responsible for the research is affiliated. Recording, storage, and safekeeping of that material are the responsibility of the individual responsible for the project. Should a researcher change institutions, the individual responsible for the project may make available a photocopy of part or all of the records, a copy of the existing electronic information, a photocopy of the data-collection notebooks, or aliquots of available biological or chemical materials, provided such sharing is necessary. A Material Transfer Agreement must be signed for all human biological samples (blood, serum, DNA, tissues etc). When the change involves the person responsible for the research, the director of the Centre will take responsibility for supervising this process.”
Therefore, normally all the data and material related to a project belongs to the institution, and should remain there. However when many stakeholders have an interest in the data, the issue of ownership may become more complex and may need to be explicitly agreed at the outset of the collaboration.
You find some unexpected outliers in the data. What should you do?
Sometimes it may be not clear cut what data to include or exclude from an analysis, and it may even make more sense scientifically – and be a more trustworthy analysis if it excludes certain outliers. If a researcher does decide to exclude outliers it is essential that these data, together with the reasons for excluding them are clearly stated in the report of the study.
How do you decide which statistical test to apply?
There are many different statistical tests and even though some experiments require a specific statistical analysis, others can potentially use several. You should always decide a priori on the statistical analysis most suitable for your study. If in doubt, get expert statistical advice before beginning the study. This will ensure you design your experiment in a way that will generate the most reliable results.
Publication of negative results
The PRBB Code of Good Scientific Practice states that: “In clinical studies and certain epidemiological studies it is equally necessary to publish negative results or results that differ from those predicted in the research project.”
Indeed, when human subjects have taken part of a study it would be unethical not to publish the results of that study, whatever they are, since that would mean their effort and time has been of no value. Also, negative findings in clinical studies can have important clinical implications in themselves and also have a great impact on safety and effectiveness and impact future potential trials.
One could also argue that not only clinical or epidemiological studies, but any kind of study – be it experimental or computational, with human samples or animal organisms – should be published. Negative results are as important as positive results. They may not be relevant to the hypothesis you were testing, but may be relevant for future studies, or for researchers working on other subjects.
With respect to this, the PRBB Code also says: “Failure to publish results or excessive delay in publishing them may constitute a serious offence relating to misuse of resources. Publication of results is an ethical imperative for clinical studies in which human subjects have participated.”
Can you find 7 differences between A and B? Is it ever ethical to use photoshop software on images? If so, what changes are permissible?
In figure B, all seven upper lines from the WT have had been made darker – while the controls (lower bars) have not been modified.
This image modification is unacceptable, because it changes the meaning of the results. Generally speaking, some changes can be made to images, to make them clearer, but some things should be taken into account when modifying images.
Here’s a series of tips for a good image handling:
- Whenever possible, keep the original image.
- Save all modified files with the change in the name to the file.
- Selective enhancement (making changes only to one part of the image) are not permissible. Changes made to the whole image (e.g. brightness adjustment) are OK in principle as long as no element of the image disappears.
- Any software used to handle images or any other changes made should be made clear in figure legends.
It is also a good idea to look at the guidelines of the journals where you are planning to submit your manuscript.
Remember you can always consult the PRBB Code of Good Scientific Practice here!