Does the best fit line for a Bradford assay standard solution pass through the X and Y axis at 0?

Question

I had did a bradford assay with 5 diff standards namely 0.01mg/ml, 0.02mg/ml, 0.03mg/ml , 0.04mg/ml and 0.05mg/ml and have measured the absorbance at 595nm agaisnt a blank. The pts that i had plot does not really give me a prefect st line graph, can anyone explain it? Also the most imp qn is whether the line should pass thru x-y axis at 0, I feel it should as I had measured the abs against a blank.

thanks bellerophon for the answer, so do you mean that the line would definately pass thru the x y axis at 0?
when I try to plot my graph, it is simply impossible to get a plot with a line thru zero.

Here are my readings
0.01mg/ml0.304
0.02mg/ml0.335
0.03mg/ml0.344
0.04mg/ml0.350
0.05mg/ml0.356

Answer 1

Welcome to the beautiful world of experiments.

When you do a standard curve you always fit the data by linear regression to an equation of the form y=ax+b.
In theory for spectroscopy you should have b=0 (Beer's Law doesn't have an intercept). But there are always experimental errors so you fit the data to the general equation for a line.

So you will have have a straight line which actually might not pass through all of the points but is representative of the function (can't draw it here, e.g. look at http://csmres.jmu.edu/biology/Bio480/spring2004/group4/plantexp/RESULT23.gif though the make it fit to y=ax, or look at http://www.msu.edu/course/lbs/145/luckie/inquiriesF2004/group/page4.htm)

Even though you make standards, they will never fall perfectly on the same line. You have many different factors introducing errors and thus making the data deviate from a perfect line. However there are criteria to judge whether the deviation can be tolerated or not and ways to minimize errors (and thus improve the quality of your curve).
If R^2<0.79 your data are worthless and you need to repeat your experiment (R is a statistical function and if you are using a programme like Excel to do the fit it also calculates R^2).
The closer the value of R^2 to 1 the better the quality of your curve. (Good standard curves should have R^2>0.90)

If you get a really bad fit, then you should make sure that you prepare the standards carefully so that the error in the true concetration is minimal. You should also make sure that you are within the linear range of the method (stay below A=1)
Use more points (usually 9 are minimum for good for linear regression but sometimes you can't make so many samples; here you can).
The Bradford assay has some peculiarities of its own. E.g. the colour changes with time and thus the value of A changes as well. So you should set a defined time period between mixing the reagents and measuring. E.g. 3 min. Then you need to make sure that for all your standards and later on the samples you keep the same time frame. You can't do a curve using 3 min and measure samples that stayed for 10 min.
If you have several samples to measure e.g. when you are measuring the standards, you will lose time measuring the samples, so e.g. if you prepare all almost at the same time, until you reach the moment to measure the absorbance of sample #4 you will have exceeded the 3 min time frame since its preparation. That's why in such cases you should prepare the samples with a delay with respect to each other (e.g. 30 sec or 1min) in order to be sure that you measure each sample at the appropriate time.

Finally, for a good curve, always start measuring from the less concentrated solution since traces that are left in the cuvette will introduce a smaller error than if you started with more concentrated solutions.


[edit] Your line will not pass through the begining of the axes. I tried to force it to fit to y=ax and it is a disaster.
You have to fit it to y=ax+b and you'll get
y= 1.19x+0.3021, with R^2=0.8486
To be honest the value R^2 is not good enough for a good standard curve. I'd try to get better quality data (I've already told you how). Looking at the plot, the first data point is really messing your line. If you reject that point you get y=0.69x+0.3221 and R^2=0.9888 which is very good.
I must point out that you ought to have samples with absorbances that span a bigger region-all your values are between 0.300 and 0.350. The highest accuracy is at the region 0.2-0.8, so I am sure you can make 9 standard solutions or even more so that their absorbances are ranging from 0.2 to 0.8.
Personally I wouldn't trust this set of data.(I am also a bit puzzled by the extremely small change in the absorbance with the change in protein concentration; something is fishy)

Answer 2

Bradford Assay Graph