• Create Account

Submit your paper to J Biol Methods today!

# minimum replicate for qpcr?

15 replies to this topic

### #1 hianghao

hianghao

Veteran

• Active Members
• 111 posts
0
Neutral

Posted 13 April 2011 - 01:52 AM

What is the minimum number of "biological replicate" and "number of reaction replicate for each biological replicate" that is required for qPCR? 2 biological replicates, each 2/3 reaction replicates will be enough or not?

### #2 Trof

Trof

Brain on a stick

• Global Moderators
• 1,400 posts
138
Excellent

Posted 13 April 2011 - 10:06 PM

I'm not sure if it's required, but usualy 3 replicates (both biological and reaction) are a good choice, because if one of the replicates flies off you can discard it and use remaining two. If you have only two replicates and they differ you can't tell which one is right and which one is off.
But when your two biological replicates show the same results, maybe you don't need to do the third.

Our country has a serious deficiency in lighthouses. I assume the main reason is that we have no sea.

I never trust anything that can't be doubted.

'Normal' is a dryer setting. - Elizabeth Moon

### #3 hianghao

hianghao

Veteran

• Active Members
• 111 posts
0
Neutral

Posted 14 April 2011 - 01:14 AM

I'm not sure if it's required, but usualy 3 replicates (both biological and reaction) are a good choice, because if one of the replicates flies off you can discard it and use remaining two. If you have only two replicates and they differ you can't tell which one is right and which one is off.
But when your two biological replicates show the same results, maybe you don't need to do the third.

I see... Is this how replicates work? Sometimes in experiment, one replicate may deviate far from the other two. So should we discard that one only and get the mean of the other 2, or repeat another replicate to replace it or use the original 3 replicates? Because doesn't it sound like manipulating the result if we discard the one with odd/non-desired result?

### #4 Trof

Trof

Brain on a stick

• Global Moderators
• 1,400 posts
138
Excellent

Posted 14 April 2011 - 03:05 AM

Well, intention is to discard the odd, never the undesired. When in group of three one is far from other two, then it's more likely that something went wrong there, you may actually like the wrong one better, but you still should discard it. Keeping it with other two and making mean of them all will shift the value and bring more error to your calculations. If you ask whether you should repeat it, well ideally yes. But that could mean repeat whole bunch of samples, that are compared with the affected one. But if your other two replicates are close to each other I don't see much point in that, apart from having triplicates "to the letter". Sometimes mean of duplicates (with low variability) is sufficient so making a reliable duplicate from a triplicate wouldn't actually matter.

Our country has a serious deficiency in lighthouses. I assume the main reason is that we have no sea.

I never trust anything that can't be doubted.

'Normal' is a dryer setting. - Elizabeth Moon

### #5 hianghao

hianghao

Veteran

• Active Members
• 111 posts
0
Neutral

Posted 15 April 2011 - 02:17 AM

Well, intention is to discard the odd, never the undesired. When in group of three one is far from other two, then it's more likely that something went wrong there, you may actually like the wrong one better, but you still should discard it. Keeping it with other two and making mean of them all will shift the value and bring more error to your calculations. If you ask whether you should repeat it, well ideally yes. But that could mean repeat whole bunch of samples, that are compared with the affected one. But if your other two replicates are close to each other I don't see much point in that, apart from having triplicates "to the letter". Sometimes mean of duplicates (with low variability) is sufficient so making a reliable duplicate from a triplicate wouldn't actually matter.

Now i got it! Thanks! Another question:
I was told that ref gene must be run concurrently with target gene reaction because efficiency of every reaction each time is different. This means that both ref gene and target genes must have same Ta. Is this true? I have 8 target genes and 1 ref gene and i found it difficult to make their Ta close. They ranged from 50-55C.

### #6 Trof

Trof

Brain on a stick

• Global Moderators
• 1,400 posts
138
Excellent

Posted 19 April 2011 - 12:49 AM

You don't need to run all genes on one plate if you're doing relative quantification. What you must on the other hand is to run all samples on one plate/run. Because you compare the samples with each other. You don't compare genes. Any difference in efficiency between genes would be canceled out.
So no, you don't need to have same Ta.

Our country has a serious deficiency in lighthouses. I assume the main reason is that we have no sea.

I never trust anything that can't be doubted.

'Normal' is a dryer setting. - Elizabeth Moon

### #7 Prep!

Prep!

Am I me???!!!

• Active Members
• 517 posts
4
Neutral

Posted 19 April 2011 - 01:31 AM

I agree with trof, though all the time if that appears and you keep doing it its not a good practice... you may have to improve on your skills... also such samples are generally termed out of trend (OOT)
Support bacteria - They are the only culture some people have!!!
Cheers!!!

### #8 hianghao

hianghao

Veteran

• Active Members
• 111 posts
0
Neutral

Posted 20 April 2011 - 04:12 PM

You don't need to run all genes on one plate if you're doing relative quantification. What you must on the other hand is to run all samples on one plate/run. Because you compare the samples with each other. You don't compare genes. Any difference in efficiency between genes would be canceled out.
So no, you don't need to have same Ta.

That means when running all samples in the same plate, i dont have to include a ref gene in it? I was told that "they must be put together (ref and target gene) in the same run because at the end we will compare the conc of target with conc of ref. Since each run has diff efficiency, it is not comparable."

### #9 Trof

Trof

Brain on a stick

• Global Moderators
• 1,400 posts
138
Excellent

Posted 20 April 2011 - 08:48 PM

That means when running all samples in the same plate, i dont have to include a ref gene in it? I was told that "they must be put together (ref and target gene) in the same run because at the end we will compare the conc of target with conc of ref. Since each run has diff efficiency, it is not comparable."

Are you doing delta-delta Ct (or efficiency corrected Pfaffl version) or something else? Are you doing relative quantification? If yes, then you compare samples not genes, as I wrote, you can have each gene on different plate.

Our country has a serious deficiency in lighthouses. I assume the main reason is that we have no sea.

I never trust anything that can't be doubted.

'Normal' is a dryer setting. - Elizabeth Moon

### #10 hianghao

hianghao

Veteran

• Active Members
• 111 posts
0
Neutral

Posted 21 April 2011 - 03:10 AM

That means when running all samples in the same plate, i dont have to include a ref gene in it? I was told that "they must be put together (ref and target gene) in the same run because at the end we will compare the conc of target with conc of ref. Since each run has diff efficiency, it is not comparable."

Are you doing delta-delta Ct (or efficiency corrected Pfaffl version) or something else? Are you doing relative quantification? If yes, then you compare samples not genes, as I wrote, you can have each gene on different plate.

i m doing relative quantification. Not sure which method to use as from what i read from qiagen qpcr manual, it mentions that the calculation method depends on whether pcr efficiencies are similar. If they are, i'll use delta-delta ct. Else, i'll convert the ct of treatment sample and convert it to conc based on standard curve, normalise target gene, then compare the expression with that of calibrator sample (dn't know the methods name).
I am studying expression of diff genes after 9 different types of treatments with diff xenobiotics and compare it with unexposed as calibrator sample. So it means that i had to put 3 (biological reps)X 3 (rxn reps) x 10 (treatment + calibrator) in each plate? What if i got more treatment and they can't fit into the plate? Do i have to do no RTase ctrl if i'd conducted DNAse treatment?
Must the conditions (eg. primer conc and template conc, dilution rate and conc used to prepare standard curve) be the same for each target gene?

### #11 Trof

Trof

Brain on a stick

• Global Moderators
• 1,400 posts
138
Excellent

Posted 21 April 2011 - 05:49 AM

It's always better to calculate with efficiencies. But you don't need to do a standard curve method to get relative quantity in that case, use Pfaffl equation, it's like delta-delta, but with efficiency (Pfaffl, M. (2001) A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Research 29 (9): 00-06).
If you have too many samples, and they don't fit to one plate you can have more runs, but you must include the same calibrator sample on every plate and normalise to the one on the same plate. This way you can have more samples.

Our country has a serious deficiency in lighthouses. I assume the main reason is that we have no sea.

I never trust anything that can't be doubted.

'Normal' is a dryer setting. - Elizabeth Moon

### #12 hianghao

hianghao

Veteran

• Active Members
• 111 posts
0
Neutral

Posted 24 April 2011 - 06:49 AM

It's always better to calculate with efficiencies. But you don't need to do a standard curve method to get relative quantity in that case, use Pfaffl equation, it's like delta-delta, but with efficiency (Pfaffl, M. (2001) A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Research 29 (9): 00-06).
If you have too many samples, and they don't fit to one plate you can have more runs, but you must include the same calibrator sample on every plate and normalise to the one on the same plate. This way you can have more samples.

I've read that paper. Crossing point mentioned in that paper= threshold cycle and ctrl sample=calibrator sample? For this model, we still need to normalise the target expression with ref expression. Still, the ref gene doesn't have to be in the same plate with target?

I am conducting 1 step qPCR. IS this model suitable for me? Because it said "...1 step RT-PCR models are not applicable..." I am not sure whether it means that this model is not suitable for 1 step or it simply meant that 2 step is more efficient than 1 step.

Edited by hianghao, 24 April 2011 - 07:25 AM.

### #13 Trof

Trof

Brain on a stick

• Global Moderators
• 1,400 posts
138
Excellent

Posted 25 April 2011 - 11:58 PM

Crossing point mentioned in that paper= threshold cycle and ctrl sample=calibrator sample? For this model, we still need to normalise the target expression with ref expression. Still, the ref gene doesn't have to be in the same plate with target?

Yes, Cp = Ct and control = calibrator. No, you still don't need to have all genes in one plate, that's same for any relative quantification. You doesn't even need to run dilution series (for calculating efficiencies) on the same plate, they can be run separate.

I am conducting 1 step qPCR. IS this model suitable for me? Because it said "...1 step RT-PCR models are not applicable..." I am not sure whether it means that this model is not suitable for 1 step or it simply meant that 2 step is more efficient than 1 step.

It says it's not applicable for their experiment with different RT efficiencies. It only says that in 1-step there is always bigger variability due to different RT conditions.

Our country has a serious deficiency in lighthouses. I assume the main reason is that we have no sea.

I never trust anything that can't be doubted.

'Normal' is a dryer setting. - Elizabeth Moon

### #14 hianghao

hianghao

Veteran

• Active Members
• 111 posts
0
Neutral

Posted 27 April 2011 - 09:45 PM

Crossing point mentioned in that paper= threshold cycle and ctrl sample=calibrator sample? For this model, we still need to normalise the target expression with ref expression. Still, the ref gene doesn't have to be in the same plate with target?

Yes, Cp = Ct and control = calibrator. No, you still don't need to have all genes in one plate, that's same for any relative quantification. You doesn't even need to run dilution series (for calculating efficiencies) on the same plate, they can be run separate.

I am conducting 1 step qPCR. IS this model suitable for me? Because it said "...1 step RT-PCR models are not applicable..." I am not sure whether it means that this model is not suitable for 1 step or it simply meant that 2 step is more efficient than 1 step.

It says it's not applicable for their experiment with different RT efficiencies. It only says that in 1-step there is always bigger variability due to different RT conditions.

### #15 Baars01

Baars01

Enthusiast

• Active Members
• 45 posts
2
Neutral

Posted 04 May 2011 - 07:27 AM

You don't need to run all genes on one plate if you're doing relative quantification. What you must on the other hand is to run all samples on one plate/run. Because you compare the samples with each other. You don't compare genes. Any difference in efficiency between genes would be canceled out.
So no, you don't need to have same Ta.

In theory, at least, this shouldn't matter if you normalise your results with a housekeeping gene which remains constant.