Experimental Conditions As Knowledge

Question

[Yogesh Pandit]

Hi,

For gene expression experiments, how important are the experimental conditions (as knowledge)?

For example, if I look for experiments done on lung cancer there will be a lot of them. Many of them will be done under different conditions. So how important do u think are these conditions to compare the different lung cancer experiments?

How can the derived knowledge be used?

Any kind of comments would be great. Thanks

[User 59]

Do you think you could add a bit more detail about what you're interested in? It sounds like you are thinking about performing some kind of meta-analysis or looking to integrate or contrast experiments. It's hard to me more specific in an answer without knowing what you're hoping to achieve.

Answer 1

[User 59]

Knowledge of the conditions is essential to the understanding and interpreting of the experiment, which is why we have things like MIAME to ensure that data that is put online in a repository has sufficiently detailed information to allow reanalysis.

To paraphrase what is on the MIAME page:

The six most critical elements contributing towards MIAME are:

1. The raw data for each hybridisation (e.g., CEL or GPR files)
2. The final processed (normalised) data for the set of hybridisations in the experiment (study) (e.g., the gene expression data matrix used to draw the conclusions from the study)
3. The essential sample annotation including experimental factors and their values (e.g., compound and dose in a dose response experiment)
4. The experimental design including sample data relationships (e.g., which raw data file relates to which sample, which hybridisations are technical, which are biological replicates)
5. Sufficient annotation of the array (e.g., gene identifiers, genomic coordinates, probe oligonucleotide sequences or reference commercial array catalog number)
6. The essential laboratory and data processing protocols (e.g., what normalisation method has been used to obtain the final processed data)

If these are captured within an ontological framework, they allow you to work out which samples should be, or could be, similar to other samples in other experiments.

[Yogesh Pandit]

How correct would it be to compare experiments on the basis of such data? These are some physical conditions, what would the comparison mean biologically?

[User 59]

Need more information than that to answer the question though! What do you mean by physical conditions?

Answer 2

[Michael]

I would like to add two points to Daniels answer mentioning MIAME:

1. It has been often said (e.g. in the MIAME publication) and I totally agree : Experimental annotation is so essential that a sensible analysis is not only hampered but totally impossible without it. This is, however, true for any experimental science I believe.
2. The MGED-ontology (it's debatable if it's complexity is beneficial) had been designed to describe in particular microarray experiments. It can be used with MAGE-TAB or MAGE-ML (MGED abandoned the overly complex XML document model they designed for annotating microarray data and replaced it with the tab-separated non-format MAGE-TAB) used with the major microarray repositories, eg. GEO, ArrayExpress.

How valuable, or just more complicated the annotation and submission process has become by using these tools, how reproducible and sufficiently annotated the experimental conditions are which can be found in the public repositories, and how much use and re-evaluation of the public microarray datasets has actually been performed based on these annotations is another story.

Answer 3

[Pierre Lindenbaum]

If you're looking after a way to describe the conditions of an experiment, you should have a look at the already existing ontologies. e.g:

• http://obi-ontology.org/page/Main_Page
• http://bioportal.bioontology.org/ontologies/45155
• http://bioportal.bioontology.org/ontologies/45155
• (...)