Hi everybody!
I was checking some books and papers related to biostatistics and I found a no clear difference between biological replicates, replicates and repetitions (especially, between the two first terms). Here are the definitions that are more clear for me:
Biological replicates are parallel measurements of biologically distinct samples that capture random biological variation, which may itself be a subject of study or a noise source. (For example, the number of animals in a experiment).
Replicates involves running the same study on different subjects but identical conditions. For example, if a I wanna know the effect of three differente temperatures on seaweed growth and I repeat ALL the experiment two more times, i have 3 replicates)
Repetition is when you take different measurements during the same experiment. For example, if I have three temperatures and i wanna know their effect on seaweed growth, in each treatment I going to have 4 repetitions, that means, 4 culture vessels in which the seaweed grows.
In my opinion, repetition and biological replicate (as i stated before) are similar. I think in most papers authors use replicates but they do not say if they really do the same experiment several times, and, instead, they are doing repetitions. Most of the time, doing replicates are more expensive.
What is your opinion? What is better for statistical analysis?
Thanks!
Yes, I agree with Rajiv.
I perform intidisciplinary resarch with Chemists (I am Biologist). In orther to understand each other we need to talk about "Biological Replicate" and "Technical Replicate".
In Chemistry, in order to measure any parameter, typically X number of technical replicates are measured. For example, we want to determine the concentration of Nitrogen in a given sample, they perform 3 times the same analysis on the same sample and the result will be the average of that 3 results.
A biological replicate is an individual of the same group in an experiment. For example, we want to test of drought on Pinus sylvestris. We need two groups of samples, Control and Drought, and within each group we need "n" biological replicates to perform the statistics, so "n" individuals under drought conditions and "n" individuals representing the control group (note that number of individuals can vary between groups! not necessarily they need to have the same number of replicates). Typically among biologists we only talk about "replicates" and not "biological replicates".
Note also that the "Technical Replicates" cannot be used for statistical purposes, so we cannot use the variability among that technical replicates for statistical inference. As I mentioned above, for statistical purposes we need the biological replicates which represent the variability of the natural population.
Hope this can help as well!
Have a good day!
Dear Jose
In statistical literature such confusion exists, as researchers from different fields sometime use different terminology. However, in statistics, particularly the aspect of "Experimental Design" where our purpose is mainly to compare the effects of some treatments , we use the words as follows.
In statistics, The replicates mean distinct samples that capture random variation. This word, when written by biologist, sometime, they wrote Biological replicates. I presume.
Now, repetition, if you see in experimentation parlance, then it is simply repetition the same treatments under identical condition.
See, this may be of some use.
Yes, I agree with Rajiv.
I perform intidisciplinary resarch with Chemists (I am Biologist). In orther to understand each other we need to talk about "Biological Replicate" and "Technical Replicate".
In Chemistry, in order to measure any parameter, typically X number of technical replicates are measured. For example, we want to determine the concentration of Nitrogen in a given sample, they perform 3 times the same analysis on the same sample and the result will be the average of that 3 results.
A biological replicate is an individual of the same group in an experiment. For example, we want to test of drought on Pinus sylvestris. We need two groups of samples, Control and Drought, and within each group we need "n" biological replicates to perform the statistics, so "n" individuals under drought conditions and "n" individuals representing the control group (note that number of individuals can vary between groups! not necessarily they need to have the same number of replicates). Typically among biologists we only talk about "replicates" and not "biological replicates".
Note also that the "Technical Replicates" cannot be used for statistical purposes, so we cannot use the variability among that technical replicates for statistical inference. As I mentioned above, for statistical purposes we need the biological replicates which represent the variability of the natural population.
Hope this can help as well!
Have a good day!
Thanks Albert! I agree with you. Technical Replicates cannot be used for statistical analysis. They are more related (but not the same) to pseudoreplication. Thanks for comment. Now my ideas are more clear! :D
A very good question!
I would encourage you to always confirm with your collaborators what they mean by these terms.
One notion of "repetition" that might be missing here is that associated with "repeated measures" -- i.e., experiments where there is interest in the effect of the passage of time between the repeated measurements on the same biological "unit". Very common: e.g., Baseline, day 1, day 3, day 7, and day 30 measurements on patients in clinical trials.
Also, watch out for terms like duplicates, triplicates, etc. Very common in lab experiments where a "single result" will actually be the mean or median of a set of "duplicate" assays (same biological sample run on same machines, same operators, etc.)
This is why there can be a greater risk of misunderstanding between experts than among relatively inexperienced researchers. At times, we are all guilty of assuming that our usage of a term is the common or obvious one.
We add to that the complexity and clarity that we have when we do sampling natural ocean waters (with phytoplankton, chlorides, metals, etc.) replicas got to meet statistical test structures? Or do we get samples that do not represent the variability of the system? Can we get statistics from samples taken from nature replicas?
Replicas means sampling from design and populations in laboratory conditions, with absolute control over the population exogenous variables. Otherwise these variables would understand by the endogenous populations. And that is fatal because we want to know the variations of populations caused by the assumption that the units that constitute the system under study can and should (thermodynamically) have differences.
If I have 30 health subjects and 30 with some disease, it will be correct do two biological replicates for each group? Considering, this aspect, it will be accep for publication ? (also we will do the techinical replicates).
Thank you for help me.
Vazimeire NASCIMENTO Oliveria
If you have 30 healthy subjects (let's consider them the Control) and 30 with a disease (those ones would be the Treatment). You finally have 30 replicates for each group.
So, your experimental design would be:
1 Categorical Factor with 2 levels/groups (Control vs. Treatment)
30 Biological Replicates each group.
I am not sure if I understood well your question. You would like to analyze only 2 subjects from each group having 30? For statistical purposes, as higher is the number of replicates, as better you will detect small significant changes between the groups.
So, do not go for two. Even you can still perform certain analyses with two replicates. It will bring a lot of concern. So, push as high as possible but never go below than 3.
In my case, for my purposes (metabolomics studies of plants), I like to analyze at least 6 biological replicates (ideally around 10).
Hope this helps!
Dear Valzimeire: Could you send us a diagram with you design? I have the same Albert´s questions. And then I´ll could send my answer about this topic with a bibliographic references and experience comments
In fact, I forgot a important detail :
30 subjects Healthy Control) and 30 with a disease (principal group)
I will do a pool of samples from each group. 2 pool from fifteen samples: health control group;
and 2 pool from fifteen samples: principal group
My doubt is:it will be correct performe these two biological replicates (two pools) for each group? It will be accep for publication ? If I can´t do 3 biological replicates.
Well... This experimental design you are proposing does not have much statistical power and the results obtained comparing two groups could be very questionable.
Think that most of inference statistics analyses (e.g. ANOVA) need to compute the "within variability" and the "between variability" to calculate the statistic-F (for ANOVA). So, how do you calcuclate the variability with only two samples?
I am from the field of ecology and as a reviewer, I would tell you that I would not trust the results coming from an experiment with just 2 replicates. I would reject that paper since the conclusions would likely not strongly supported by the data.
Why do you have to pool the samples?
Cannot you pool 10 + 10 + 10 for each of the groups?
I can’t agree more with comments of Albert Rivas-Uback. In fact, from my opinion, the words “Repetition” almost meaningless, so as the “noise”！ All individual, population, community and ecosystem are different, what we can (and should do) is compare variability within and among subjects hierarchically. Moreover, we should never overlook variability within subject as a “noise” (e.g. Individuals), which is the resource of evolution.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis