graph for an ELISA

[deebird24]

Hey folks,

 

If any of you have ever done an ELISA and have spectro readings like these, can you tell me if you would do a LOG graph for them , or how you do them on excel?

 

Concentration Absorbance

3000 3.148

1500 1.627

750 0.88

375 0.453

187.5 0.257

93.75 0.146

46.88 0.095

23.44 0.076

11.72 0.057

5.86 0.05

2.93 0.047

1.46 0.042

 

As you can see the concentration was double dilution, and I felt the only way I could fit this into a coherent graph was to convert it into LOG, but my lecturer has said, why did you that?

[CharonY]

A log or semi-log plot is easy to do, but I do not really see that it is necessary here. Just by looking at the values it is apparent that you are outside the dynamic range of the assay in certain areas of your graph (or assuming that you used absorbance reading in a standard photometer the dynamic range of the instruments is to blame).

 

In any case, to make such a plot just calculate the logs using the =log function in excel or simply scale the axis to a log.

[ewmon]

Was your lecturer questioning that you used the log of the Absorbance to plot the points on regular graph paper or semi-log paper? If so, then I agree with your lecturer. Instead, keep the original Absorbance values and plot them on log-log paper.

 

I also agree with CharonY. Because you're halving the concentrations serially, simply double an Absorbance and compare it to the previous Absorbance. There's a certain spot where the readings obviously begin to "go wrong". You can also see this in the log-log plot — some points will fall on a straight line, but others will deviate significantly. If it happened at the other end of the dilutions, then it would be a matrix effect.

[deebird24]

1353580400[/url]' post='714604']

Was your lecturer questioning that you used the log of the Absorbance to plot the points on regular graph paper or semi-log paper? If so, then I agree with your lecturer. Instead, keep the original Absorbance values and plot them on log-log paper.

 

I also agree with CharonY. Because you're halving the concentrations serially, simply double an Absorbance and compare it to the previous Absorbance. There's a certain spot where the readings obviously begin to "go wrong". You can also see this in the log-log plot — some points will fall on a straight line, but others will deviate significantly. If it happened at the other end of the dilutions, then it would be a matrix effect.

 

We had to submit electronically, so I used Excel and performed a log-log graph. I am not questioning the absorbance results, I needed help from people to see how they would scale these values, would they too use a LOG graph? Given that the values range from 2 to 3000. I thought I done the right thing, I thought that data points should be spread evenly on a graph, but if you just plot the [C] as say, 3000, 1500, 1000, 500 and 0. - then what you get is a big clump of data points before the 500 mark on the X axis. This was why I converted to LOG. I just thought graphs were supposed to be spread out, but as I said originally, my tutor has asked me WHY I done log-log graph.

[CharonY]

For a calibration curve you would generally only use data points corresponding to dynamic range of the measurement. The reason is that in most cases you want to perform a linear regression analysis as a measure of the quality. For the most part these type of analyses follow linear or exponential functions and as such at most semi-logs are used.

Log-log is more appropriate for power law relationships.

In this particular case a log-log curve does not tell you much except (as ewmon pointed out) showing the part where the linear relationship fails and hence the calibration curve is useless. Not necessarily unimportant, but you can estimate that by eye, too.

[ewmon]

Not to get off track, CharonY, but I began my biochem career in a small clinical lab where we used the term "calibration curve". Then I moved into MNC biopharma where my terminology was corrected to "standard curve". Does this somehow jive with your experience? Thanks.

[CharonY]

Both are commonly used interchangeably, but in my experience calibration curve is more used in settings where assays are being established or developed or when instrument tweaking is involved. "Standard curve" is more often used when a suite of (more or less) established analyses are being performed.

 

This may due to the more flexible way to establish assays and the use of "standard" or "normalization" could be slightly confusing. But for the most part it is semantics.

Edited November 26, 2012 by CharonY