# Please help me on this ELISA analysis - Cannot figure out to transform OD into concentration (Oct/18/2008 )

Hello, I need help. I think I almost give up

I performed ELISA test few days ago, and now trying to get the result by analyzing it. I am kinda new for ELISA; this is the second time I did and I got problem that I cannot solve

In previous research, I used GraphPad Prism and it worked well. This time, I tried to do the same but the software could not draw a graphic for me, also could not yield "unknown" samples from the equation or interpolation. I have no idea on what I can do next.

A friend of mine suggested to use Excel, work on linear regression, however my curve is not linear at all. When I tried to apply the linear trendline, the R^2 is 0.6... and when I crosschecked to kit's standards, almost all results are wrong. Another friend suggested me to use polynomial #2 but it didn't work either.

Manually, I'm drawing a standard curve with semi-log paper and pencil now. I know this will generate bigger error but I cannot think anything else.

If there's someone here would like to help me, I will appreciate very much. Please send me PM and I'll send my raw data (OD). I already arranged it nicely, everything is there, but I just cannot finish it properly. Please, I need your help. Thank you in advance. I'm waiting...

If you used graphpad prism correctly and it is unable to draw a graph ther is someting wrong with your data.

I would draw the graph with semi log paper and pencil and if it looks like a normal graph then it is a normal graph and you can take it for granted that in graphpad something is not OK.

If not provide the data and I'm willing to look into it. If possible the raw data without any correction.

Hello, Gerard!

Thanks for your reply

GraphPad can draw the standard curve (from the "known"), but it couldn't show me the result for the "unknown" aka my sample. For my calculation/data manipulation, I found there's a little part of my sample that have values which are above the highest standard's concentration. I tried to exclude these samples, but still it didn't work for me

So, I started with paper and pencil and have my nonlinear curve, very nonlinear. At this point, I couldn't generate the equation for applying the "unknown"

I expect that you could tell me what I should do.

Please check your PM inbox, I sent you the raw data (in Excel).

Thanks in advance!

What type of ELISA are we talking about? Is it a direct/sandwich ELISA or a competitive?

If it is a direct or sandwich ELISA you should really just use the linear portion of the standard curve - when the curve starts to level off your accuracy goes out the window (not to mention when you extrapolate). This you can do in Excel by just removing the high standard points that do not fall on a straight line (your r^2 should be > 0,9) and using the equation given for the trend line to calculate the concentrations of your unknowns.

If you are doing a competitive ELISA you probably need to use the logistic fit types you can find (somewhere) in Graph Pad - 4 and 5 point logistic fits mimic the shape of a competitive assay standard curve the best. I can't really remember how to use them off the top of my head (I don't have GraphPad installed currently), but as I recall you have to convert your data into logarithmic form first.

You can graph by hand and it should be OK.

You can only interpret the points in the linear portion of the dose response curve.

The immunoassay curve shape should be sigmoidal depending upon how you graph it.

Sounds like you are now on the right track.

I use tablecuve software.

I performed ELISA test few days ago, and now trying to get the result by analyzing it. I am kinda new for ELISA; this is the second time I did and I got problem that I cannot solve

In previous research, I used GraphPad Prism and it worked well. This time, I tried to do the same but the software could not draw a graphic for me, also could not yield "unknown" samples from the equation or interpolation. I have no idea on what I can do next.

A friend of mine suggested to use Excel, work on linear regression, however my curve is not linear at all. When I tried to apply the linear trendline, the R^2 is 0.6... and when I crosschecked to kit's standards, almost all results are wrong. Another friend suggested me to use polynomial #2 but it didn't work either.

Manually, I'm drawing a standard curve with semi-log paper and pencil now. I know this will generate bigger error but I cannot think anything else.

If there's someone here would like to help me, I will appreciate very much. Please send me PM and I'll send my raw data (OD). I already arranged it nicely, everything is there, but I just cannot finish it properly. Please, I need your help. Thank you in advance. I'm waiting...

I use excel for my elisas as well - i use a kit (BD biosciences, and they provide cytokine standards which are linear on a log-log scale. I plot OD vs. concentration in excel as a SCATTER graph (you don't need to make the axis logarithmic), and fit a POWER trendline to it. this will give you an equation with a variation of OD = constant1 x 10^ (constant2 x concentration). so then you re-work the equation to get in terms of concentration = 10^ (LOG(OD)-LOG(constant 2))/constant 1. plug in your OD from your samples, and you'll get your concentration.

hope that makes sense.