The bumbling biochemist’s pipetting thumb is sore because with technical REPLICATES it’s more, more, more! You know how on medical dramas when they run a test and get a weird result or they find drugs in a patient’s pee and confront the patient and the patient says something like “you must have made a mistake – run it again!” and the doctor says something like “I already did. 3 times. And they all came back positive”? Well, if the doctor really did run the sample 3 times he would have performed technical replicates.
Technical replicates are one of 2 main types of experimental replicates: TECHNICAL REPLICATES test the same sample multiple times to account for variation in measuring whereas BIOLOGICAL REPLICATES test different samples. And both of these are different from INDEPENDENT EXPERIMENTS, where you test different samples on different days with fresh setups, etc.
Independent experiments can account for things like “there was something in the water” or more commonly when it comes to biochemistry there was something left out of the water! (it’s really easy to accidentally forget to add things so you want to develop systems like moving tubes to a different rack after you add them, or checking off things you’ve added on a piece of paper (but word of caution, it takes a while to develop habits so in the beginning days you can confuse yourself more because you might add something but forget to cross it off or move the tube so then you think you haven’t added it when you have! Oh – and another thing that helps -I recently discovered these “cluster tubes” that are like deep well blocks but with removable wells which are great for when you don’t have a full block’s-worth.
Each of these types of “double-checking” have value but in different ways. To help illustrate the difference, let’s look at an example. Types of replicates are often explained in terms of patients or lab rats – e.g. say you treat 10 people with a drug that’s supposed to lower blood pressure and then you measure the treated people’s blood pressure. If you measure the same person’s blood pressure 10 times (maybe you thought the machines was acting weird or something – or it took the person a while to relax) those would be technical replicates (they tell you about how reliable the measuring is and variation within the sample). If you measure each person’s blood pressure those would be biological replicates (these will tell you about differences between how people respond). And if you repeated the experiment with a different group of people, that would be an independent experiment (this tells you about how representative of the wider population that first group was)
Say 1 of the people responded really well to the treatment but the other 9 didn’t. If you measured that 1 person’s blood pressure 5 times but everyone else’s just once and then you took the average it’d seem like the drug worked a lot better than it actually did. So instead when you average, you average the averages of the technical replicates (so you’d take the average of that strong-responder so it doesn’t skew the results). So, if you have 10 people, your “n” is 10 regardless of how many measurements you make.
I don’t work with people (well, I do work with people but I don’t do research on them) or animals – but I do work with a lot of replicates. Whether they’re different protein preps (biological replicates) or – what my pipetting thumb’s sore from today – tests of the same protein prep but repeated multiple times to account for things like pipetting differences, etc. (technical replicates). For some of these, it’s the first time I’m testing the samples (so no independent experiments yet) but for others I’ve done it before so it is an independent experiment.
In addition to protein activity assays (experiments to measure things), a common place you might come across similar concepts in a molecular biology lab is with molecular cloning. This is where we stick a gene into a circular piece of DNA called a plasmid that we can get bacteria to host for us. We have to check that the gene actually got in there. And one way we can do this is with an analytical (aka diagnostic) restriction enzyme digest.
Restriction enzymes (restriction endonucleases) are DNA scissors that recognize and cut at specific DNA sequences. If you take plasmid DNA that you think has your gene and you add a restriction enzyme that cuts a sequence that’s in the gene – the presence or absence of said gene leads to different bands being seen! more here: http://bit.ly/30Npa8o
You want to check that your gene got in ok because it doesn’t always (the digest isn’t definitive but it can be a good quick check before you send the “hits” for sequencing to be sure all’s ok). When you do cloning you do this thing called transformation where, after engineering the plasmid to contain your insert you convince bacteria to take that plasmid in. A common way to do this is by bathing chemically weakened “competent cells” with the plasmid and briefly heating them (in a “heat shock” step to take it in). The plasmid also has an antibiotic resistance gene so that only bacteria that took in the plasmid can survive when you plate it on food spiked with the corresponding antibiotic.
So you let them grow on plates and form colonies and then you pick a few colonies, grow them in some liquid broth to make more of the plasmid DNA, and then you isolate that plasmid DNA – often with alkaline lysis methods like “miniprep” kits. more here: http://bit.ly/2MNGJ50Then you can test them. Each colony represents a different clone (the bacteria grow by splitting so each colony is a clump of identical bacteria but the bacteria in different colonies might be different, so you test multiple colonies).
The other day in our lab a couple of my labmates were jokingly trying to “one-up” each other by bragging about how well their cloning went. So let’s pretend that they did an actual competition where they were comparing whose worked better by picking 10 colonies from each of their plates and running a restriction digest on them to see who gets the most “positives.” Each clone would be a “biological replicate” and if they ran 3 digests on each clone those would each be “technical replicates”
For experiments like this where you have a lot of replicates to do, you often make a “master mix” – more on those here: http://bit.ly/2m5nzfV
But basically you make a mixture of everything but the unique thing – then you add that same mixture to different unique things. In this case, each reaction will need:
- plasmid DNA (this is the unique part)
- restriction enzyme
- buffer (pH-stabilized salt water with any cofactors like metals the enzyme needs to work)
- water to dilute things to the final reaction volume you want
So you can make a master mix of the enzyme, buffer, and water and add some of that same mix to the different plasmids. If you made a mistake making the master mix, all the samples will be skewed (e.g. if you forgot to add the enzyme all your tubes will turn up “negative” even in your gene’s in there – which is why you should run a “positive control” that you know should get cut to check that all is a-okay). And this is one of those places where you can see the value of the independent experiment! But if you just make a mistake in one of the technical replicates, only that sample would be affected.
When figuring out “who won” you can’t count each technical replicate as a population-increaser or else 1 colony that got 3 “positives” would count more than 1 colony that got 2 positives because there was some pipetting error for one of them.
The types of replicates I was working with today involve proteins and here’s an idea of the sort of thing replicates mean in this context.
- e.g. same protein prep, repeat experiment in parallel or take multiple samples from the same reaction
- variation in measurement – consistent pipetting (was there an air bubble, did you forget to add something)
- variation in sample – was the sample mixed well? did you happen to take a pipetful that was super full of stuff?
- more technical replicates -> better estimation of the mean but does not change sample size
- how representative is your sample?
- different people or animal subjects, different cell lines, different protein preps
- are differences you saw in one sample really real? Are they just background variation
- human error?
- equipment error?
Sorry today’s post got more rambley and awkward than usual – I was doing a lot of replicates today to prepare for a thesis committee progress meeting and I want to make sure I have solid data to show! So I’m exhausted but s a new lab book day which I always love – and this one makes an even dozen!