someone who knows more about scientific rigor, study design, methodology and such could chime in on the topic of "scientific validity".
Yes, I completely agree, scientific validity is the key. I am actually at a pretty good position to discuss about it. During my PhD I did lifespan studies in mice, developped
statistics concerning the number of animals to use (my first post ever;), got training and the diploma to lead
in vivo experiments in mice and rats, and read quite a lot about the topic.
The key things I was taught are:
control everything you can to avoid every possible bias (no animal stress; age, gender, strain, light; pathogens in air, food, water, handling; modify one thing at a time and do all controls), use a minimum number of animals but not too few, same for experiments (ie just the right ones, not too many persons on the same animals), and most importantly:
prepare everything! Plan all you can and test all you can: you don't want to have any bug nor difficulty of interpretations otherwise the experiment loses its value. I have about 5 kilos of documents about the subject;)
I personnally consider Richard Miller as one of the most recognized scientists on the subject.
Here are 10 advices he and Nancy Nadon give (also available in pubmed.org/10795715). One good example of what happens in case of pathogens is his
mouse lifespan study on methionine restriction: difficult to interpret; it also shows that pathogens are difficuly to avoid since it happens to the best people.
However, R Miller and the ITP take some distance about the 'scholar' rules: rather than testing one specific strain of mice they test of
'standardized heterogeneous' mice, eg in order to avoid studying the effect of some drug on one particular disease of one particular strain. The perfect test would be to test each strain separately, but all things considered, their heterogeneous mice seems to be the best compromise. Same for the age at which to start the intervention: a birth? young/medium/late adulthood? The best is of course to try everything but at some point compromises must be done.
In our case (MPrize @ home),
- it is not possible to control everything. It is not possible to standardize everything (cage, light, handling), More importantly, it is not possible to be a specific pathogen free environment. In a lab, the big danger with pathogens is that they don't create smooth heterogeneity but rather typically increase the death rate of the unlucky group that gets MHV (mouse hepatitis virus) independently of which treatment it has. Or they increase the death rate of all groups and you end up studying MHV, not aging. That's why the idea of MPrize @ home initially shoked the scholar guy that I was. But the more I considered it, the more I took some distance with the strict rules I had learned, and found some positive in the negative effects.
- distributed lifespans is a force. As for heterogeneous mice, we don't want to study disease/frailty that happen under some particular condition (eg 12h light 12h dark everyday, no stress, non disease) that are not representative of the population. We can not avoid pathogens, but we can distribute animals in as many places as possible so that 1) we don't put all our eggs in the same basket (ie we have less risk of something condeming the whole experiment) and 2) in our aging conditions we include colds and any disease that affect individuals. If a disease is known to affect all mice at home, then it is normal to include it in our search for life extension. If it affect one third of the population, our sampling should be representative. Here, we see that the number of homes is important for statistical power, (not only the number of mice; this is already the case for the number of cages, even in a traditional lab).
- we should plan our study carefully in order to take advantage of our conditions. To summarize the latter paragraph, because we can not be pathogen-free, we should try to have many homes and few animals per home rather than few homes and many animals per home. Having 2 animals only per cage is however not a good idea because if one dies the other will die permaturely, an effect that I don't want to study particularly. Having 4 animals per cage, without replacement in case of death, seems the good compromise to me. Also, to increase statistical power, we can try to have control cages next to treatment cages (if one cage gets a disease, the other is likely to get it too), as long as we have plexiglas cages (ie animals don't throw food and treatment to the neighboring cage) and as long as we have a methodology not to separate treated from control animals (ie never change both cages simultaneously, have a trick to give treatmentA to cageA and treatmentB to cageB).
We should go on considering everything. We might have to do compromises (eg testing for many diseases costs a lot). If other people start being interested I'll start having a few animals at home just to test every procedure and make sure everything is easy and clearly establish when there are risks (I'm used to traditional labs, not mice @ home). I would be pleased if other/more experienced people could interact to help us find caveats. Since it is partly a non-conventional approach I would expect reactions from experienced people and new ideas from non-experienced people.
To end my posts, a few disparate things:
- "undergrad research project" : why not, but for efficient use of ressources i think that distributed environments (people over the internet) should lead to distributed projects, whereas undergrads doing a project in a conventional lab probably had better work on an approach for which the lab is good at (cells, one-place animal studies)
- "supporting SENS", synergism: see MVivo, Methuselah Lifespan tests . However if I mentor the methionine restriction project with MFURI I indeed won't be able to have mice at home (to avoid contamination)
- example of research project: http://mfoundation.o...p...wpost&t=779 Perhaps also something with TauRX if it is sufficiently ready.