1
$\begingroup$

Experimental design: Cells are grown, counted, and seeded in 96-well plates. Cells recover 24 hours. Cells are treated with different doses of a drug, 3x per treatment. Cell survival read via MTT.

This is repeated twice, on two additional times.

In other words, on three independent days, we seed out this plate and measure it one day afterwards (measurement kills the cells):

A B C D
1 1 µM 2 µM 5 µM 10 µM
2 1 µM 2 µM 5 µM 10 µM
3 1 µM 2 µM 5 µM 10 µM

Each Dosage (columns) is measured in triplicates (rows). This is repeated three times, so in the end, each dosage has a total of nine datapoints.

Does this experiment have 3 biological replicates, each of which has 3 technical replicates or is there another way to group this?

My inclination is to use a mixed-level model where each of the plates is a level in a random variable.

$\endgroup$
6
  • $\begingroup$ I don't think your description of the experiment is very clear. $\endgroup$
    – Bryan Krause
    Dec 12, 2022 at 21:40
  • 2
    $\begingroup$ @ Bryan Krause: I disagree, I understand his setup very clearly, it‘s a standard procedure in cell culture. Also, this question should not have been closed, since these matters aren‘t trivial and their impact is often underestimated $\endgroup$
    – markur
    May 15 at 22:39
  • $\begingroup$ @markur Disagree, it's still unclear. It may be standard, but it's unclear. $\endgroup$
    – Bryan Krause
    May 16 at 14:48
  • $\begingroup$ @BryanKrause please specify what exactly is not clear. I have ideas how to improve this post, but I don‘t want to change things that you already understand. $\endgroup$
    – markur
    May 16 at 15:54
  • $\begingroup$ @markur How the replications are administered: what is measured in the same wells and what is measured in different wells. Total N is important too. $\endgroup$
    – Bryan Krause
    May 16 at 16:22

1 Answer 1

1
$\begingroup$

It's a bit complicated, because your cells at each replicate are single source, so biological replicates are not truly replicates - but it gets very difficult to escape that when you consider that most established cell lines are from a single source. If you want to do it as close to properly as you can, you should get up 3 separate vials of your cells and maintain/use them individually.

In your case you can consider that different repeats are biological replicates (so you have initial with 2 repeats? so 3 replicates total(?)) - your wording is unclear here, I interpreted it to mean you have done the experiment twice, not an initial with two additional repeats, but the next line says "3 biological replicates". These give you information about the biological variation in your cells over time.

The replicates at each repeat are technical replicates and provide information about how well you are able to do things like pipetting or making dilutions.

$\endgroup$
5
  • $\begingroup$ The experiment was done three times. Regarding cell lines, yes, all the cells came from the same lot of the same ATCC catalogue entry (albeit different "break-outs" from frozen stocks). So, there are actually zero biological replicates? $\endgroup$
    – Bryan
    Dec 13, 2022 at 16:14
  • 1
    $\begingroup$ @Bryan it's a philosophical debate as to whether they are biological or not. Most would consider them to be biological replicates, because if you reductio ad absurdum the argument, it makes it impossible to ever have biological replicates in a cell line. $\endgroup$
    – bob1
    Dec 13, 2022 at 22:17
  • $\begingroup$ It is not philosophical. Technical replicates measure technical variance while keeping biological variance as low as possible. Hence, you measure technical replicates on the same day. Biological variance is introduced by prolonged culturing (react to differences of FCS, culture flasks, Selection by splitting, etc.). To display biological variance, you measure at different timepoints and minimize technical variance by taking the mean of tech replicates. Even if the cells have the same background, some cell lines do have great internal time-dependent variance, like the ones I work on. $\endgroup$
    – markur
    May 15 at 22:38
  • $\begingroup$ Also, it‘s important to never mix technical and biological variance. If you measure triplicates on three different days, show technical variance by plotting three separated groups, each group consisting of the three technical replicates, (not one group with nine datapoints). To show biological variance, take the mean of the triplicates and display the three means, each mean measured on a different day (not as a ‚time series‘, just on a different random day). If you mix technical and biological variance (n=9) , that‘s pseudo-replication and and bad practice. $\endgroup$
    – markur
    May 15 at 22:49
  • $\begingroup$ Source: Harvey Motulsky (the CEO of Graphpad Prism), Intuitive Biostatistics $\endgroup$
    – markur
    May 15 at 22:54

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .