Most fields have general accepted practices. For example, I believe NCBI accepts microarray data, NGS data, even SNP data. I think only certain formats are allowed so you can search the database.

You present a common problem. I know large biomarker studies often try to define data collection exactly between their collaborators, since differing practices can confound the results. Ultimately I think it's up to each field to make such decisions.

Thank you for responding!  You mentioned two things to which I would like to call to attention; collaborators and standards of different fields. 

What I am after is what a common measurement standard for collaborations between different fields would look like and what would happen if experts with the ability to improve  standards are frustrated enough to make it their problem to change things. 

Would that be a simple problem or would they face resistance?  Would it be worth the effort and would they be applauded/rewarded for making it their responsibility or would it lead to a difference in opinion among collaborators that ruins the project?  If we put the data side by side with existing information standards, would it lead to anything other than elicit responses like, "oh, that's beautiful" or "oh, that's just description"?  Is this attempt at improving things justifiable to warrant more funding?

Should this discussion be about concepts or is the problem so diverse that we have to discuss actual examples?

There are groups that are making important contributions to this, for example MIQE (http://miqe-press.gene-quantification.info).  They are part of a larger group (MIBBI, Minimum Information for Biological and Biomedical Investigations, http://mibbi.sourceforge.net/portal.shtml) that is attempting to standardize reporting of data gathered using a variety of techniques.  MIQE is specifically working on standardized performance and reporting of qPCR, which is often done incorrectly due to its apparent simplicity because, in fact, it is not simple at all.  MIBBI and groups like MIQE are making important contributions to improving the situation.  

Hi Dany,

Thank you for the links.  I looked up the information and goals of the MIBBI (in particular The Mouse Phenotype Database Integration Consortium project), which sounds very similar to other integrative mapping projects I’ve seen (e.g., Mabee et al.,). I respect the difficulty and nuances of the MIBBI project, in particular because they have to deal with non-organismal samples, that is, samples that do not maintain the wholeness of organisms.  However, there is a statement in the MIBBI Nature Biotech paper that confuses me.  They state,

“These various analyses make two things plain: first, that there are standout priority areas for the MIBBI Foundry (for example, the uniform description of an organism); and second, that there are many niche areas where little or no collaborative activity is required (for example, the process of mouse phenotyping)…”

This doesn’t make sense to me because my expectation of a “uniform description of an organism” is computing an organism; externalizing our mental models of developmental/evolutionary sequences, fleshed out with our knowledge.  To me, mouse phenotyping cannot be anything other than a collaborative effort because the number of things I wish to know about a developmental sequence (e.g., form, hierarchical connectivity, physiology, genetics, material parameters, etc…) is more than that contained by current disciplinary boundaries. 

I share the attitude expressed in the Levin website.  For example: 

“While data grow exponentially, true insight into shape generation and repair is significantly impaired because bioinformatics is focused on gene sequences but not applicable to analyses of shape. Thus, several fields are stymied by a lack of conceptual and computerized tools to link mechanistic understanding of molecular signals with behavior of the patterning systems they encode.”

So, how do we coordinate our vision for what should go into a database that contains all we wish to know that will not conflict as you move across levels of organization?  Should analyses be organized as specific durations to manage the information content?  Which durations are appropriate for in vivo hemangioblast generation?  What’s lacking that we can’t simply jump from an analysis of a planarian/fruit fly/worm/zebrafish and move on to a mouse system? 

Here is a concluding statement from Henkelman in "Systems Biology through Mouse Imaging Centers: Experience and New Directions" (2010).

Greater Need for Systems-Biology Modeling to Sort Out the Genome-to-Phenotype Relationship With ever-growing databases such as the Mouse Genome Informatics, the Mouse Phenome Database, and the Gene Expression Atlas, there will be a high demand for engineers, mathematicians, physicists, and computer scientists to work with geneticists, developmental biologists, physiologists, and clinicians to understand the connection between genomes and phenotypes for the mouse and consequently to learn more about human development and the genetic basis of disease. There will be plenty of challenges for all of us!"

These ideas were reiterated in a recent webcast on the Human Placenta Project (2014).  They also highlight all the different types of information that different experts are interested in knowing about.  Moreover, everyone wants an integrated version but what does that look like?

http://videocast.nih.gov/summary.asp?Live=14257&bhcp=1

Perhaps the leader in such atlasing efforts is the BRAIN initiative (a link to a lecture by Sejnowski is given below).  So, what can we learn from what they're doing in Neuroscience that we can apply to morphology? 

Given that these issues continue to persist despite meetings of experts for identifying problems, is there something missing in our approach?  Whose problem is this?

http://videocast.nih.gov/summary.asp?Live=14130&bhcp=1

The Levin lab web page discusses our main interest, which is morphogenesis, the generation of shape during development and regeneration.  And I agree - lacking a standardized way of describing shapes inhibits interdisciplinary collaboration; in fact, it inhibits intradisciplinary comparisons.  Planform  ( http://planform.daniel-lobo.com) is one of the lab's attempts to standardize the description of planarian shape so that experiments can be compared.  In that case, the model is for one person to establish the standard and everyone agrees to participate.  We hope it will work.  

Am I correct that it is descriptions of 3D shape that we are talking about standardizing? There are a lot of issues under discussion here: how to describe shape as well as how to get everyone, in all fields,  to describe shape in the same way.  The former is the more easily addressed, the latter is about human nature, which is way over my head.   Metrology, the study of measurement, is out there.  There are methods for describing shape, geometric morphometrics being the one I know best.  I'm sure the topologists could teach us about shape descriptions.  But I think it has to come from a recognized authority.  NIST and the international body of meteorologists might have some very useful information on deciding on a standard and making everyone stick to it.  Ultimately, though, I believe it will be up to the Natures, Sciences, Cells, and the equivalent players in other fields, to make the editorial decision that they will not accept non-standardized descriptions.  If we must do it to publish, we will do it.  Given that Nature will publish graphs showing the s.e.m. instead of the standard deviation, I think it is an uphill battle.    

With regards morphogenesis:

That video demonstration of Planform is “awesome example” (audio?)!    :)

My view of morphogenesis is a modern version of Goethe’s morphology, which is defined as “’the science of the form (Gestalt), formation (Bildung) and transformation (Umbildung) of organic beings;’ Morphology in Goethe’s view, then is the study of the internal laws of biological organization [Lenoir, 1987, p. 24].  For Burdach the aim of morphology was to study ‘‘the meaning and significance of organic forms..., the laws of formation...and...the process[es] by which...creative Nature brings forth organization, the mode of origin, and progress of a structure...’’ [Nyhart, 1995, p.39]. (Opitz, American Journal of Medical Genetics).  “His aim, as he described it, was "to do nothing less than to present to the physical eye, step by step, a detailed, graphic, orderly version of what I had previously presented to the inner eye conceptually and in words alone…”(Miller).  This, to me, is the externalization of reasoning, which Johnson-Laird summarizes as, “reasoning is a simulation of the world fleshed out with our knowledge, not a formal rearrangement of the logical skeletons of sentences.”

With regards descriptions of 3D shape:

Yes, we are talking, in part, about standardizing shape but I take shape to mean a lot.  That is, these are issues of atlasing and mapping but what makes maps useful is an economic representation of what it is we wish to know since reality encompasses a LOT of information.  Imagine, for example, a life-size version of the world.  Moreover, since we’re talking about collaboration, the trick is to make a representation that can satisfy all who investigate, those who focus on different aspects of the same problem. 

For example, I am interested in describing shape, but I want to focus on nuclear shape, which can be considered as “open”/“closed”/“fractal globule”/in combination that can be further developed into polymer models that fill the entire volume.  So, if a given cell is in a particular combination of states, it will be expected to be simultaneous with specific given –omics signatures.  The goal is to standardize this entire thing, including shape of tissues/organs, because it will be inclusive and comprehensive.  One can then imagine/predict, given an experimental –omic signature, what the 3D state of the nucleus was at the time of measurement and vice-versa.  Any discrepancies would lie in how well we’ve calculated the inverse problem (given –omics signature, what is the shape of the nuclei/genomic organization?), how much our current knowledge of nuclear states compares with respect to future representations, and other issues such as heterogeneity/phase of cells at time of collection, number of cells, etc…

This broadly defined view of the inverse problem is generic to many types of issues we face about measurement and understanding biological processes.  Using Planform as example, we can answer “how did we arrive at a phenotype with 4 heads?” by simply looking up by the type of perturbation, when we implemented treatment, etc., and fitting the information in context of other things we know.  

A formalized statement of the inverse problem can be given in abduction, which can be stated in simple form as: “Given P,  and Q - > P, infer Q”, which is easier to do once we have an ideal type (P) for comparison.  I see MIBBI’s “ideal mouse” (The Mouse Phenotype Database Integration Consortium) as another example but one planarian doesn’t project to simply one “ideal mouse”.  There are more cells, life-cyles are longer, more developmental transitions (?) etc…

With regards recognized authority:

I agree with you that having a recognized authority announcing standards would make a difference but it’s not clear who that authority should be for projects that integrate across disciplines.  As a biologist, I would be irritated if I found out metrologists are announcing how I should take measurements, particularly since their awareness of the vast and nuanced behaviors of living samples is limited.  I think it’s critical that we have examples to look at so we can decide on our own but which examples?

For example, Alberts et al., (PNAS) recently published a perspective dealing with future science, which addresses overlapping issues as above.  Others have stated related concerns but I think the reputations of the authors (and our perception of current research environment) make the force of the message different this time. They also warn, “Our immediate goal has been to stimulate debate of the issues that concern us and the changes we propose. The task cannot be left to a self-appointed subset of senior scientists like ourselves or to the leaders of the NIH who are known to be considering many of these same problems (6). We therefore encourage academic institutions, scientific societies, funding organizations, and other interested parties to organize discussions, national and regional, with a wide range of relevant constituencies.” 

Moreover, although I respect those who worked to meet the standards of publication in Nature, Science and Cell, their standards are not always perfectly thought out (cf., Vidal and Newman on ridiculogram). 

With regards connectomes, Bargmann and Marder (Nature Methods, 2013) state their sensibility on what our goals should be: 

“But how should we approach building models of large networks without generating models that are as difficult to understand as the biological systems that motivated and inspired them?... We are in the midst of a fascinating international debate about whether it is the right time to embark on a ‘big science’ project to monitor and model large brain regions… Big science works best when the goals of a project are well-defined and when the out­comes can be easily recognized. Both were true about the human genome project, but neither is true, yet, about large-scale attempts to understand the brain… That said, the largest challenge we face in future attempts to understand the dynamics of large circuits is not in collecting the data: what is most needed are new methods that allow our human brains to understand what we find. Humans are notoriously bad at understanding multiple nonlinear processes, although we excel at pattern recognition. Somehow, we have to turn the enormous data sets that are already starting to be generated into a form we can analyze and think about. Otherwise, we will be doomed to creating a machine that will understand the human brain better than we can!” 

I have to disagree with their (often too commonly shared) position that the biggest challenge is “not in collecting the data”.  We don’t know what the biggest challenges are, yet, although figuring out ways to understand the data is in contention for being the hardest problem. 

I love this conversation on the what we want to know and what we can measure of biological samples...

To think that it's even more complex...I guess that's why it's 21st century science.

http://mendelspod.com/podcast/are-we-driving-innovation-without-quality-pete-kissinger-purdue