Hello, can anyone help me with determining the statistics of how many embryos I need to inject to state that the phenotype or rescue I am seeing from my morpholino or mRNA injection is statistically significant (taking into consideration the penetrance and # of embryos that survive)? Also, from your experience, what percentage of your control morpholino embryos survive the microinjection after 24 hours? Thank you!
Hi,
Morpholino injections cause around 10 -25 % death in 24 hour embryos depending on morpholino concentration and the embryo batch so make sure to have WT uninjected embryos as well as control morpholino injected embryos as controls.
I would suggest injecting approximately 250- 300 embryos to establish a statistical significance. You can do this in 3 different days and pool the data in the end. The more the better of course but if you can't then this number should be enough. Also make sure that your morpholino concentration is not too high as this would be lethal and cause off target activation (p53 activation and therefore apoptosis).
Hope this helps.
Hi Alexandros, Thanks for the answer. So what do you base your statistical significance on? For example if you have 225-270 (90% of 250-300 embryos) that survived after 24 hpf morpholino injection, then if more than half of them has a phenotype, can it be reported as significant? Or you cannot really say anything about statistical significance based on the phenotype, and you can just make a table showing the number of embryos w/ or w/o a phenotype for all groups (uninjected, control, morpholino)?
Hi,
yes you can talk about statistical significance. You should discuss with your supervisor or with a colleague on how to define each phenotype and if they are consistent. For example you can divide them into Severe, Moderate and Mild. This may not be very reliable but it is up to your judgement on how to score each phenotype (that's why WT controls are very important). I am sending a graph of a typical data presentation on phenotypes. It contains the percentage of each phenotype and the total number of surviving embryos at the bottom. (It should also include WT embryos but not in this graph). A table should also be presented.
Hi Richard,
A statistical test you can use to analyse proportions can be found in the following paper: Taillard ED et al (2008) European Journal of Operational Research. v185,p1336
A drawback is that it can only be used when you have two "outcomes". For example, you can compare the effects of your morpholino against a control morpholino if you classify the embryos only into "affected" and "unaffected", or "survived" and "dead", etc.
Another option is to measure whatever it is that you are actually studying. For example, if you were looking at somitogenesis, you could count the number of somites in control and experimental embryos after 24hpf. If you were looking at cell migration, you could measure the distance that the cells had migrated, and so on. (You need to be careful to keep controls and experimentals in exactly the same temperature/conditions to make sure they are the same age.) This allows to you use other statistical tests (t-test, etc) and also to analyse variation.
So, it will depend on your experiment and what you are actually trying to assess as to the best way to analyse your embryos. I completely agree with Alexandros that it is essential to look at a number of categories (dead, severe phenotype, mild phenotype etc) to get a good idea of what is going on - realistically, you are going to get more variation than just "affected vs unaffected". You can then decide to pool the different categories and use the stat test above to test "affected vs unaffected" proportions to confirm the effect of the morpholino if you think it would be useful, and/or to measure the effect in other ways.
Reading back, I'm not sure if this necessarily answers your question ... but hopefully at least some of it will be useful!
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics