Hi QPCR experts,
I hope for some expert advise from you. My problem is following: I am testing gene expression in monocytes derived from peripheral blood of patients with various immune pathologies (by immunomagnetic separation, Dynabeads). I always was used to low total RNA yields (RNeasy mini kit, Qiagen) around 20 ug/ul in a volume of 40 ul. But now, when analyzing data from a specific group of patients suffering from sepsis (mentiong cause it may play a role...?) I am getting even lower yields: 0.5 - 4 ng/ul (total volume 40 ul)!!!!! All the procedures are same as in other groups of patients, my hands are the same... These samples are pretty rare so I did run QPCR with cDNA made from this low RNA. My question is, can I use the results = Cts (if I even get some signal) from my gene of interest (40 < Ct < 50)? I am asking if the data is reliable, if there is not some bias caused by extremely low starting RNA concentration... :-( I'd prefer to exlude 1/3 of my samples though it means to decrease the group VERY much, than to work with and rely on biased data and make wrong conclusions...
Thank you for any suggestion.
Paja
0
3 replies to this topic
#2
tea-test
-
- Active Members
-
- 168 posts
Veteran
17
Good
Good
Posted 19 April 2011 - 12:14 AM
yes, if your target in your sample is present in very low copy numbers your results can be biased.
There is no normal distribution anymore between your samples anymore but a so-called poisson distribution.
Let me explain by the words of some statistic book i have read some time ago:
In general, people often equalize "randomly distributed" with "equally distributed" but this is not the same.
Imagine if you are baking a cake, you put in 20 raisins and you mix the dough. if you afterwards divide the cake in 20 pieces, do you think you will find exactly 1 raisin in every piece? for sure not, maybe you have 3 in one piece and 0 in the other. The same with your target copies in your sample tube if they are very sparse.
But if you put in 200 raisins, the error rate will be much smaller like 10 +/- 2 in every piece (just arbitrary numbers).
I hope, this helped.
There is no normal distribution anymore between your samples anymore but a so-called poisson distribution.
Let me explain by the words of some statistic book i have read some time ago:
In general, people often equalize "randomly distributed" with "equally distributed" but this is not the same.
Imagine if you are baking a cake, you put in 20 raisins and you mix the dough. if you afterwards divide the cake in 20 pieces, do you think you will find exactly 1 raisin in every piece? for sure not, maybe you have 3 in one piece and 0 in the other. The same with your target copies in your sample tube if they are very sparse.
But if you put in 200 raisins, the error rate will be much smaller like 10 +/- 2 in every piece (just arbitrary numbers).
I hope, this helped.
tea-test: The artist formerly known as Ned Land
#3
Trof
-
- Global Moderators
-
- 943 posts
Brain on a stick
66
Excellent
Excellent
Posted 19 April 2011 - 01:07 AM
Thanks tea-test, I always knew it as "small numbers problem" but now I Know how is that properly called.
In my experience Ct values higher than 35 are generaly unreliable.
In my experience Ct values higher than 35 are generaly unreliable.
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.
I never trust anything that can't be doubted.
#4
krusty
-
- Members
-
- 3 posts
member
0
Neutral
Neutral
Posted 22 April 2011 - 03:35 AM
Hy guys
You are talking about RNA, cDNA etc...
Do you have an indication of the expected cDNA yield after a cycle of cDNA synthesis from RNA? I know it depends on RNA quality and quantity, but I only wanted to know if it is correct to expect a 100% transcription efficiency in the case of a "perfect" RNA sample, or if this percentage is lower. And, in the case of lower quality and low-concentrated RNA (c.a. 15 ng RNA), is it likely to expect very low efficiency? (e.g. 10% or lower)
Any help is appreciated, any reference as well.
Thanks in advance
Bests
K
You are talking about RNA, cDNA etc...
Do you have an indication of the expected cDNA yield after a cycle of cDNA synthesis from RNA? I know it depends on RNA quality and quantity, but I only wanted to know if it is correct to expect a 100% transcription efficiency in the case of a "perfect" RNA sample, or if this percentage is lower. And, in the case of lower quality and low-concentrated RNA (c.a. 15 ng RNA), is it likely to expect very low efficiency? (e.g. 10% or lower)
Any help is appreciated, any reference as well.
Thanks in advance
Bests
K
