I am performing several TNFa ELISAs using an R&D kit. My best results are curve heights are between 1.0 and 1.2 OD units. With the naked eye I see colour change in the three most highly concentrated standard wells and 3 or occasionally 4 standards fall on the linear part of the curve. I think my plates & reagents are OK because some of my samples have higher ODs than the standard curve.
It seems that the similar ODs of the 5 or so low concentration standards create inaccuracies reading in the lower part of the curve- so it is hard to know if the high nils seen in the attachment are real.
I'm running a field lab in a remote location, any views on steps to take before I conclude the standards need to be replaced greatly appreciated.
Hi there. The standard curve as far as it goes looks pretty good, but I see what you mean relative to the low end. It's hard to trouble-shoot this sort of thing effectively, but a couple of simple things I would try are:
increasing the time of the incubation with the colorimetric substrate (the step before adding the H2SO4). [nb I'm assuming that you have everything standardized by time…]. This time can easily be 15, 30 or 45 minutes, but it should never be 2 or 5 or "until it's dark enough"
recheck all your dilution arithmetic, particularly for your standard curve's highest concentration.
What I would NOT do is increase the concentration of any component - this will likely simply raise your background.
Looking at the data, I'd ask though whether you need to change anything - the pathogen stimulated data are all readable within the range of the assay, which in the end is what you want - if there were a lot of samples with "undetectable" then I'd look at the bottom end of the assay (actually, I'd look at your stimulation conditions), but perhaps the best advice is to ask whether you're achieving what you're setting out to.
Good Luck
G
Hi there. The standard curve as far as it goes looks pretty good, but I see what you mean relative to the low end. It's hard to trouble-shoot this sort of thing effectively, but a couple of simple things I would try are:
increasing the time of the incubation with the colorimetric substrate (the step before adding the H2SO4). [nb I'm assuming that you have everything standardized by time…]. This time can easily be 15, 30 or 45 minutes, but it should never be 2 or 5 or "until it's dark enough"
recheck all your dilution arithmetic, particularly for your standard curve's highest concentration.
What I would NOT do is increase the concentration of any component - this will likely simply raise your background.
Looking at the data, I'd ask though whether you need to change anything - the pathogen stimulated data are all readable within the range of the assay, which in the end is what you want - if there were a lot of samples with "undetectable" then I'd look at the bottom end of the assay (actually, I'd look at your stimulation conditions), but perhaps the best advice is to ask whether you're achieving what you're setting out to.
Good Luck
G
Thanks Grant, those are very helpful points. And yes I see that we've missed the highest standard (should be 1000). I'll fix that and then try with extending the incubation as you describe. Thanks again. Ayesha
Dear Ayesha,
R&D has quality control samples, with low, medium and high values that you can run along with your samples, in order to control your curve. These control samples have pre-studied values, and not only can help you check your curve performance, but also will help you control the variation of values within different assays.
Regards,
Catarina
It's a perfect curve. Also the OD's are OK. they could be an little bit higher (up to 2.0), but not more. The absorbance is the reciprocal values of the transmission. Therefor higher values only seems to be better.
I would try Grant's suggestions. If your future curves have this same shape (high asymmetry), use a 5 parameter logistic curve fitting model (5PL). If the curve shape is sigmoidal and symmetrical, use a 4PL curve fitting model.
i would say you std is not titrating after A6 B6 well, and reading A2 is not correct. I hope you are changing tips when making dilution of stds. Your higher values seems to me correct values. The other option is to manually calcuate your std curve.
The Rodbard model describes the chemistry behind best. http://osdir.com/ml/java.imagej/2004-12/msg00049.html
Curve fittings (4 or 5 parameter regression models) are normally in ELISA software packages included
hi
usually for elisa method..four or five parameter logistic or spline fits or point to point is recommended
I believe a 4PL curve requires upper/lower asymptotes, an inflection point, and slope (the 4 parameters). Using a 4PL with the curve provided above would produce poor results as there is not anything close to an upper asymptote. A 5PL can handle this asymmetry. Findlay, J. W. A., & Dillard, R. F. (2007). Appropriate calibration curve fitting in ligand binding assays. The AAPS Journal, 9(2), E260–7. doi:10.1208/aapsj0902029
- Badges
- Science method
- Similar topics
- Medicine
- Immunology
- Immunology Techniques
- Immunoassay
- ELISA