Technical replication asks this question: what kind of variation do I see that is my fault, or the error of the process, and is not connected to variability of the biological process I am studying. In the case of qPCR, technical replication such as running triplicate wells of a sample shows issues like pipetting error, or the variation in fluorescent reading by the PCR machine. 

A different kind of technical replication would be if you divided a single sample - let's say a whole leaf - into half and extracted DNA from each piece.  The difference in concentration and quality of DNA from the identical pieces only highlights a technical issue - the quality of the extraction procedure. If you see a large difference, this means your extraction procedure is inconsistent; you can not trust that whatever mistake happened here will not happen in other samples.

Biological replication, on the other hand, looks at the variability of a response from similar individuals, so that you can gauge the range of results "in real life" that have nothing to do with technical problems such as pipetting or machine error.

So for your experiment, you have to define for yourself what is an appropriate biological replicate - is it a whole plant? A single leaf? Does one contribute more  significantly to the experiment?

For example, let's say you are studying the rate of translocation of radioactive sugar from roots to leaves. If you sample 6 leaves from a single plant and treat them as biological replicates, you can only claim your results as appropriate for that ONE plant, and there is no real replication in your study. If, however, you treat 6 plants with radioactive sugar, and sample one leaf from each plant, that is a better measure of biological replication, because you are testing multiple root-to-leaf systems. But what if there is a difference in result between the top leaf and bottom leaf of a plant? If you sample a top leaf from plant#1 and a bottom leaf from plant #2, you will have a very large range of variability among your biological replicates (different plants), which gives you bad data. A good solution would be to sample top AND bottom leaves (and maybe some in the middle) from each plant, and pool top+bottom together for each plant. This evens out the in-plant variation from your biological replicates, which are the different plants. 

It all depends on the purpose of your experiment. Pooling is problematic, unless you are convinced that the samples that contribute to a pool are not on the level of biological replicates. Generally, a time point is NOT a biological replicate - this is an experimental condition which you have tested. And remember, technical replication data should not contribute to the statistical analysis of your biological data - it is merely a measurement of how much you can trust your method of data generation, and proof that your data is of good quality. 

Technical replication asks this question: what kind of variation do I see that is my fault, or the error of the process, and is not connected to variability of the biological process I am studying. In the case of qPCR, technical replication such as running triplicate wells of a sample shows issues like pipetting error, or the variation in fluorescent reading by the PCR machine. 

A different kind of technical replication would be if you divided a single sample - let's say a whole leaf - into half and extracted DNA from each piece.  The difference in concentration and quality of DNA from the identical pieces only highlights a technical issue - the quality of the extraction procedure. If you see a large difference, this means your extraction procedure is inconsistent; you can not trust that whatever mistake happened here will not happen in other samples.

Biological replication, on the other hand, looks at the variability of a response from similar individuals, so that you can gauge the range of results "in real life" that have nothing to do with technical problems such as pipetting or machine error.

So for your experiment, you have to define for yourself what is an appropriate biological replicate - is it a whole plant? A single leaf? Does one contribute more  significantly to the experiment?

For example, let's say you are studying the rate of translocation of radioactive sugar from roots to leaves. If you sample 6 leaves from a single plant and treat them as biological replicates, you can only claim your results as appropriate for that ONE plant, and there is no real replication in your study. If, however, you treat 6 plants with radioactive sugar, and sample one leaf from each plant, that is a better measure of biological replication, because you are testing multiple root-to-leaf systems. But what if there is a difference in result between the top leaf and bottom leaf of a plant? If you sample a top leaf from plant#1 and a bottom leaf from plant #2, you will have a very large range of variability among your biological replicates (different plants), which gives you bad data. A good solution would be to sample top AND bottom leaves (and maybe some in the middle) from each plant, and pool top+bottom together for each plant. This evens out the in-plant variation from your biological replicates, which are the different plants. 

It all depends on the purpose of your experiment. Pooling is problematic, unless you are convinced that the samples that contribute to a pool are not on the level of biological replicates. Generally, a time point is NOT a biological replicate - this is an experimental condition which you have tested. And remember, technical replication data should not contribute to the statistical analysis of your biological data - it is merely a measurement of how much you can trust your method of data generation, and proof that your data is of good quality. 

Thank you for that clear and concise explanation / clarification. Could find in many research papers where samples are pooled and similar strategy of pooling is followed for validation of sequencing results with qPCR.

Hi, Sujatha I completely agree with Amy about sample pooling among biological replicates.Adding few points pooling of the samples is often done in order to avoid variability among the biological replicates when you checking gene expression via microarray or profiling. Pooling of the sample is usually done when the sample size per group is too much for example in your case lets assume 12 plants and you have 2 different groups and you studying the whole gene expression pattern among them so it is not possible to profile each sample as it will be expensive and plus data normalization will be difficult so in this case you can pool four plant samples each group to get one so you will end up with 4 samples each group this reduces the sequencing or profiling cost and sample variation among biological replicates.Pooling should be done among samples belonging from one group.in your case, as Amy said about the leaves it is a good way to reduce variability  but pooling top and bottom leaves will be not  a good if you are doing a tissue specific study because you won't be specifically able to reach a conclusion that the overall changes are contributed more by the top leave or bottom leave, in that case, your pooled samples should be the same tissue always. but it completely depends upon your study.  

2) At times sample variation can also be seen even if you collect it from the same location suppose bottom leaves in all plants and top leaves in all plant still because the way a person collects the sample, change in the collection methods, reagents used all can contribute towards  variation among the replicates and at times few of the genes are in circadian Rythm so to overcome these  trying collecting the sample at same time points and  pooling will help you  to avoid all your data variation.

3) Data Interpretation: During validation of results you can use the same pooled samples and run a real-time in triplicate(for each gene under investigation). In PCR first step take the average of Ct values of  (Technical replicates) to minimize variation among them do that in all 4 samples after this in next step once you normalize the data with your endogenous control to get delta Ct values for each sample (n=4) per group just take the average and find the st dev  calculate the fold mean fold st dev . 

 biological replicates can be made from each and step before cDNA synthesis

but if possible, make replicates different groups, different individuals, different sample from a same individual. 

Hi,

Thanks a lot for the replies and suggestions. Am studying tissue specific expression of genes. If 3-4 plants are selected as biological replications is it necessary that all the tissues should be selected from the same plants/individuals or just any 3-4 plants (different tissues will be from different plants)?. Also, due to asynchronous growth, if a particular stage is not reached, can we collect samples from any 3 plants when the stage is reached, or only from the same 3-4 plants chosen as biological replications?

Again, it depends: If it is necessary to pair the data from different tissue types, then they really should be from the same plant which is the same biological replicate. However, if the data from different tissues are not relevant to each other, then I think it's ok to use different individuals from separate experiments.  

For example, if you want to say that Gene X is expressed at 1 hour in the stem, but does not appear expressed in leaves until 3 hours, they really should be sampled from the same plant at both time points. If you do the sampling from different plants, there may be enough variation in the environment to falsely widen the gap between tissue-specific expression (too much water in one pot, or one plant under more light, etc. such that the gene is triggered differently).

On the other hand, if you want to say Gene X, which is involved in transpiration, is expressed in the stem at 1hr, and Gene Y, which is involved in formation of chlorophyll, is expressed in leaves at 3 hours, that could be done in different experiments, from different individuals, because these two tests have very little to do with one another (I assume; I am making this up).

If your experiment is designed around an organism that is not well synchronized, such as a mature plant, then I would suppose it is more important to wait for the stage to be reached and keep the same biological replicates, rather than stick to a human schedule.  So going back to GeneX, if you also want to compare expression in young roots and then flowers and seeds, you still really should use the same plant over its entire lifespan. I think it would be understandable if seeds were not available from biological replicate #1 at the exact same time as biological replicate #2 and #3. But you should look to experiments you are more familiar with than I am to answer this question.

Hi, Biological replicates usually can be described as an individual of the same group in an experiment. 

Study 1: when you say Biological replicates that mean let's assume if you working on a 4 different plant species and you are interested in studying the expression of gene X in a specific tissue. Then ideally you will collect the same tissue from 4 different plant species  (each plant species samples will be collected in triplicate or can also be increased ). So here your four different plant species are making your 4 different groups and each group further can have 3 plants (Which will be your biological replicate) for that specific species.  Then you can do a qualitative real time PCR.In this study sample of one group can't be replaced with a sample of another group as all group depicts different species in which you comparing some genes. 

Study 2: Another example just like plant can be mice based studies You have 2 groups 12 mice in each group  so in this case, you have two group one healthy and second diseased in which you are inducing some disease lets take obesity for an example and you are interested in comparing genes related to obesity among both groups. Unfortunately while checking you found out that only 10 mice developed obesity 2 couldn't so now in this case you should proceed with 10 mice and leave those 2 now you have 10 biological replicates. 

Adding few points too pooling of samples just to avoid variation the pooling should be done in the same way because each pool will serve as a biological replicate and are assumed to have equal contribution  so if bulk comes from one plant so the whole RNA will be similar to that plant but what we want is an average .

I hope it will be helpful !!!

Hi Tarun,

It was very helpful.