By Liberty Foreman, SSI Fellow and PhD Candidate, ISMB, University College London
- The keen response toward the workshop idea from all the clinical fellows. All of whom would like to attend the workshop as either a participant or an expert.
- A final itinerary for the workshop was decided. The workshop will be held for a maximum of 7 projects in September 2014. These will be selected via an application process in July. A suitable expert collaborator will then be chosen based on the requirements of the project. Both parties will then meet for a lunch on the day of the workshop, and work together on their data in the afternoon.
- The workshop project has also had some recent support from the head of my department, and the head of the statistics department at UCL.
In September 2014 I will be holding a workshop that aims to marry software developers, statisticians and the like, with clinical fellows at UCL. The aim is to promote and produce reproducible research within the medical environment, as more often than not clinical fellows do not have the time to gain expertise in coding to handle their large datasets. This report describes the aims and conclusions of the clinical fellow focus group meeting held on 14th April. The aim of this focus group was to gain insight into what clinicians would want from a workshop, and how the workshop should be carried out for maximum benefit not only for the clinicians but also UCL academics acting as experts.
Ten clinical fellows were recruited from a wide range of research areas, all of which generate large datasets. Their fields include immunology, epidemiology, proteomics, genomics and oncology. These recruits have very diverse lab settings; a few have some access to data analysts whilst others have none.
There were three facilitators present; I chaired the meeting following the agenda described in the pre-claim report. A colleague, Katherine Oliver, agreed to act as a scribe and a facilitator. Finally, my SSI fellowship supporter/ supervisor, Dr. James Brown, who helped recruit the attendants, also helped facilitate the meeting.
Information established from the attendees
Eight of the clinical fellows had basic backgrounds in data analysis. Most of these recruits were at the data collection stage, carrying out only basic analysis to aid the direction of their experiments. The general feeling was that, at present, they had enough skills to carry out the basic analysis. However, the main task of data analysis felt daunting to most. All of these recruits were extremely keen on the idea of a workshop. More enthusiasm was shown at the prospect of collaborating with computational/statistical experts and potentially gaining more customised tools or software than currently used.
The other two clinical fellows had more advanced levels, both with Masters degrees in data analysis for epidemiology and immunology. Therefore, their experiences were different from that of the other recruits. Being comfortable with Python and R and handling large datasets, both of these fellows found themselves helping out clinical colleagues. These two attendees both felt they would not require the services of a workshop. However, one showed interest in being involved as a data analyst.
There were three main tools used for data analysis; SPSS, Microsoft Excel and Graphpad Prism. The two recruits with analytical expertise used Python and R, and one other had very basic skills in R. Some concerns were raised with ‘black box’ methods, but relative comfort was expressed with the output, although most admitted to not understanding the methods.
Currently analytical collaborations begin through word of mouth with the aim of exhibiting credibility in that area of expertise in a publication or grant application. These collaborators were described as advisory, with little to no contact time. There was one example where authorship was given to a collaborator as a result of a conversation that happened a few years prior to the article. These issues show a demand for longer term collaborations with more contact time. In order to illustrate how we would like longer term collaborations to work I used my PhD project as an example; I work in direct collaboration with a clinical fellow who recruits patients and collects epidemiologic data, and a statistician who directs the research. In addition to directing the research, the statistician also use the data collected to develop their own theoretical advances. Therefore, the principle investigator is motivated to aid in my research development because he gains access to a real and growing dataset. This helped alleviate the one major concern of the clinical fellows at the focus group that the principle investigator might lose motivation to work on the project.
Existing grant application support services, the Joint Research Office (JRO) and Research Design Services (RDS), in place specifically for clinical research were discussed. These services require involvement from the project concept stage, all the way through to grant application and then publication. Those who had dealt with these services would use them again. However, issues such as cost, time of involvement and commitment to the services were raised.
The clinical fellows showed most enthusiasm for an afternoon workshop where selected clinical fellows would work on their data with a suitable expert, who has both the time and motivation for a longer term collaboration.
The timeline for the workshop has been determined as:
Jul 2014 - Deadlines for applications: a simple application form outlining the project and expertise required submitted by clinical fellows.
Jul-Aug 2014 - Review of all applications followed by selection and matching of seven projects with motivated experts.
Sep 2014 - An afternoon workshop, followed by a networking session where candidates would give 2 minute talks on their experiences of the afternoon.
Follow up events
Since the focus group meeting I have had a meeting with Prof. Tom Fearn, head of the UCL Department of Statistics and Prof. Gabriel Waksman, the director of the Institute of Structural and Molecular Biology. Both have shown enthusiasm for the project and have agreed to support the workshop and help promote the idea to principal investigators in their departments. Prof. Fearn advised that to make the recruitment of statistical experts successful, we should only include projects that contain large datasets. In addition, he also said that for a successful collaboration, the project owners would have to be willing to allow the collaborating group to use the datasets for their own theoretical development. This idea had been mentioned and accepted at the focus group meeting.
In summary, I believe that the focus group meeting and the follow up meetings have significantly helped in the design of the workshop. The workshop is also gaining momentum as more departmental heads have praised and support the idea. I am now in the process of advertising this event both on the academic side to gain experts, and the medical side to increase the number of applications.