BOF: How to shift perception of bioinformatics from supplementary work to equal expertise

[borisevichdi]

Let's face it - most of the PIs in biology are biologists or MDs, and many of them were born, when general purpose computers hadn't yet became a part of everyday life. In their days, all the "data analysis" was done using pen and paper by the biologist. Therefore, bioinformatics (as well as computational biology, statistics, and so on) are often not perceived as a separate, equally complex expertise field, but just as overblown, unnecessarily overly complicated part of statistics.

This can lead to situations, especially in small groups, such as:

• Bioinformaticians are not consulted during the experiment planning phase, leading to experiments, whose results can not be used for analysis in the end,
• Bioinformaticians are not consulted during the experiment planning phase, leading to too simple analysis plan, which PI can come up with, and our skills are not utilized to the full extent and benefit,
• Bioinformaticians are not involved for the QC, when the first results are delivered, and expensive experiments are being run, even when it would be clear from the very beginning, that something is wrong,
• Bioinformaticians are expected to run analysis and deliver results irrespectively of the data quality / analysis power / etc.,
• Bioinformaticians are not expected to be able to provide an intuition and understanding of the results, it is a biologists "privilege", while it is apparent, that data scientist analyzing the data have much deeper intuition.
• Bioinformatics is expected to be free, or at least cheap, causing insufficient funding and unrealistic time frames.
• Bioinformatic speakers are virtually non-existent at the topical conferences and it is harder to justify the international stay for bioinformaticians.
• "Owning" the data seems to give a lot of rights for the MD/biologist, whereas "owning" a script does not, and therefore bioinformaticians are incentivised to do shitty pipelines only they can use, and not creating proper packages and tools.

To change the situation, I believe that we, as bioinformaticians (and everybody else), have to assert ourselves and ensure to be heard. Here are some provocative and too ambitious goals, which give an idea about the desirable direction, IMO:

• Obligatory approval of grant proposals, right to deny any processing requests for proposals submitted without a detailed discussion with us (including grants, associated with hiring new phds/postdocs).
• Shared first co-authorship for each publication.
• Dedicated funding for improvements and development of IT infrastructure.
• Funding for outsourced data analysis / software development, similarly to when someone pays for the sequencing in BGI, someone must also expect to pay for the data processing.
• Obligatory involvement in hiring to ensure that new-coming person can be worked together, both in terms of technical stack and personality. No one will hire a biologist with skills, that does not match the lab research, but hiring say a golang person into R team is seemed completely fine somehow.
• CS conferences, funded and organized by our, biological, institutes.

As a results, all parties will benefit, because:

• PIs get a better results, avoid wasting time and money on badly designed experiments,
• We get better recognition and opportunities for career.

Now, how to achieve this paradise is a huge question, and this is what I want to discuss, if you have the same problems. And if you have not, then I encourage you even more to join the potential session, to share with others how do you ensure a proper workflow in your lab.

[borisevichdi]

Additional discussion topic:

Permanent Senior researcher officer positions for computer scientists - how to promote and where to find funding?

[borisevichdi]

Notes

Roots of the bioinformaticians problems

A brainstorm on problems and their problems was performed.

Roots of the problems, which were prioritized the highest, are:

• Individual, not team-based, evaluation of researcher results.
• Historical perception of bioinformatics as supplementary service.
• Ego of PIs.
• Highly heterogeneous background / training of the bioinformaticians, and highly heterogeneous expectations from them.
• Conservative feedback loop (publications -> grant -> publications), done as it was done 20 years ago.
• "Environmentalism" - obsession with Nature.
• Lack of professionals and consequent split of duties of PhD and Postdoc bioinformaticians.
• Not planning bioinformatics as a part of a research project, post-hoc planning of data analysis.

Additional honorable mentions:

• It is hard to assert yourself as a single researcher.
• Career opportunities are not stable / long-term.
• PIs do not always seek help for software development.
• Bioinformaticians are perceived as service providers, not equal partners.
• MSc and PhD students are perceived as free workforce.
• Strong hierarchy exists in the lab, unlike in CS. Additionally, different disciplines / research areas are seen as less/more "prestigious".

Possible solutions

The top 6 wild solutions, proposed by the participants, to solve the most problematic issues, were:

• Highly complex:
  1. Unionisation / collectivization - bioinformaticians are sparse, and they need an uniting entity, who would represent and lobby their interests.
  2. Funding agencies requirements - funding agencies must require grant proposals to include bioinformatics methods as a part of the proposal, together with costs attributed to this part of research.
• Middle complexity:
  3. Infiltration of decision makers - bioinformaticians should be more incorporated into the decision making organizations (funding agencies, reviewers, centers' top management, etc.) with veto rights, so that they can ensure that research lacking bioinformatics will not be acceptable.
  4. Marketing of bioinformatics - value and prestige of bioinformatics analysis and bioinformatician as a profession should be increased via proper marketing/advertisement of them.
  5. New Journal(s) on Bioinformatics - more journals for bioinformaticians to ease publication and break into the research feedback loop.
• Low complexity:
  6. 1:1 external co-funding for bioinformaticians - for each 1 wet lab person, hired into an institution, external funds should be provided to hire 1 bioinformatics / data management / software development employee externally (for the NNF centers, for example, by NNF), so that high level of bioinformatics is ensured and PIs would not worry about "spending" money on informaticians.

[borisevichdi]

Thanks everyone for joining the workshop! And feel free to correct me, add additional ideas and opinions, etc.