Question

What Is Most Important In A Bioinformatics Project Description?

9

Entering edit mode

13.2 years ago

Sean Davis 27k

A little hypothetical scenario:

An academic institution is setting up a bioinformatics support group. This group will service many investigators with many levels of expertise and interests, though the focus will be on genomics data analysis in human on both microarray and sequencing technologies. Since this resource will be quite limited, there is a need to gather highly relevant information about potential projects. Assuming that the original submission is going to be electronic, what are the most relevant pieces of information to gather. Examples might include:

Biological questions to be answered
Technologies employed
Experimental design including sample descriptions and number of replicates
Proposed bioinformatics plan (if one exists)
Previous and current bioinformatics expertise involved
Other related datasets (either public or local)
References to similar studies and background
Computational resources available (storage, commercial or open source software, workstations)
Type of project--pilot, followup to previous study, full publishable dataset, other

Are there examples of such details available? I am aware of MIAME and similar standards, but what I am seeking is a more "operational" set of descriptors that will allow reasonably facile characterization of projects.

project • 6.2k views

ADD COMMENT • link updated 12.7 years ago by Mary 11k • written 13.2 years ago by Sean Davis 27k

Ram · Answer 1 · 2011-10-12

You know... Oftentimes bioinformaticians working at a university are simply scientists that are good at working with computer software. So they don't like to do services, and they don't like to think of themselves as doing services at all. Now I think that what you really want is to solve biological research questions and a group of bioinformaticians and biostatisticians that will contribute towards solving those biological problems, that will think of approaches to do that and that will be creative in doing so. That might sometimes be just doing standard analyses, sometimes giving feedback on the quality of the results, sometimes indicating that you could do more with a different kind of data, and sometimes they will bring up completely new research questions. So I think you might consider to forget about a service group and to build a systems biology center instead, where bioinformaticians, biostatisticians, wetlab researchers and probably also modellers work in groups directed to specific research problems. While each of the specialists involved of course also will meet up with people from his own discipline.

If nothing else you might at least get better, more motivated bioinformaticians in that way. But you might also prevent that wetlab researchers produce data that even the best bioinformatician cannot make sense of or that your service center people don't really understand your questions and solve problems that nobody has.

If I misunderstood your question and it was really about describing the research projects, like what were the questions, what was the study design, what were the methods used, samples produced and data delivered you might want to look at the newer multiomics ISA standard (see http://www.isa-tools.org), or at tools designed to capture such studies like our own GSCF (see http://gscf.dbnp.org)

score 4 · Answer 2 · 2011-10-12

I will add three bits of info right off the top:

Statistical analysis - to be done by the support group, the researchers/customers, or both in collaboration, and which methods to employ.[?] Payment - this support won't come without real costs, so the plan to pay should be stated up-front. This may include "free for now, but plans are to move to a fee-for-service model."[?] Future projects anticipated that will incorporate these results - the support group should be informed or forewarned as to where all this is headed (one-time analysis, or the beginning of something grander or on a repeat basis)[?]

The Bioinformatics plan should exist - more grant proposals to NIH and others demand this! Perhaps a fourth item could be a simple question about these results/analysis leading to a grant proposal and if the support group will be included in the writing of that proposal, and will the proposal describe monetary support for the support group. If the support group is successful and/or needed to the point that a description of that support group is included in the Facilities section of the grant, then the grant writers surely ought to be prepared to bring the support group into the Research Plan and Budget sections of that proposal early in the writing process.

Added in edit: Authorship. Better discussed up front than at any time later.

score 2 · Answer 3 · 2011-10-13

If I could ask for something I would also ask for a curation/visualization plan for the resulting data if it is going to be public. Where will it be submitted, such as a repository or visualization strategy (dbGaP, dbSNP, GenBank, ORegAnno, custom tracks, a custom GBrowse, and so on), and whose responsibility that is. Are there curation needs beyond that--such as WikiPathways or other places specific for the type of data involved? That could be significant work for large data sets and I'd love to see it discussed beforehand.

Maybe this is further downstream than you need for your purposes. But if you can have that conversation it might be valuable in assessing the project too.