Biochem is a filthy whore

ceres

Let's see if anyone's awake.

So I need to find protein concentrations taken from random aliquots during an experiment. I have my absorbance standard curve, and I'm *pretty* sure that after several hours of staring at the damned thing it's finally right.

So I start plugging the absorbance data taken, and I'm getting negative numbers. I can't have negative concentrations. My 'b' in the y=mx+b equation is .27, and some of my absorbances (y) are less than that. The data fell within the range of data in the original standard curve (before b was worked out), and I can't retake it now.

Am I missing something or am I screwed?

tofu

Sorry, I'm not following you, what do you mean the data fell within the range of the original standard curve?

The y-intercept, b, should be 0 so there is certainly some error in your measurements, have you accounted for standard deviation?

ceres

Well, when I say it fell within the the range of data used for the curve (which was all that was required at the time), I mean that I didn't have the equation handy. The range ended up being lowest absorbance recorded --> highest absorbance recorded. I have not accounted for standard deviation because I am awful and don't know how to do that.

But what's happening is that the .27 means that for any absorbance I get that's less than .27, I get e negative x value, which is unpossible because I can't have a concentration debt. The equation itself is something that the spreadsheet program spat out at me when I made the trend line. All the data points are not perfectly in line.

I have this feeling that the answer is to retake the data, and that's not going to be possible.

ceres

Additionally I feel it's important to note that I actually have no idea what I'm talking about. I have class during my prof's office hours and she hasn't responded to my emails, and this thing is due in like 3 hours. :P

tofu

Would you mind posting the data and your spreadsheet on google docs?

ceres

https://docs.google.com/leaf?id=0B55UiEluQjIGMjVmNTg3MjEtMjA1MC00YzQyLWE5MmItMGZiNDFlMzViODFh&hl=en&authkey=CNuZx4MM

That's the spreadsheet and data for the curve. One example (experimental) data point it .2414, which is lower than the .27 and will give me a negative I have no way to get rid of.

romanqwerty

Well i'm a chemist, not a biochemist, but that data tells me that something is different between your control (calibration line, and your samples) that produces extra absorbance in addition to the analyte you're measuring.

Or, the alternative, your blank actually contained some analyte. If you have no way to retake the data, you're basically screwed... You might be able to get a rough idea by forcing the trendline through (0,0) but thats not a great method.

ceres

It's kind of messy because the aliquots aren't "clean". We're looking for protein concentration with these readings (it's a coomassie assay), but there's a lot more than protein floating around in there. At most of the stages that are giving negative x values, the aliquots are basically lysed yeast cells with the membranes and heavier junk settled out.

I'm not sure that matters, because the lowest point in the data used to generate the curve was like .23something.. it seemed like we were safe with a .2414.

tofu

The R^2 for the trendline is alright as it is, forcing to the origin would make it unusable. Some other analyte in the standards probably shifted all the absorbance values up

edit: Wow that sounds incredibly inaccurate

ceres

Well, bleh. So basically, we probably should have tried for a lesser dilution, and there's nothing I can do about it now.

This stuff is not my strong suit. I don't math good.

Thanks for the input, I appreciate it.

tofu

Sorry I can't be of more help, I'm a chemist and work with trace metal values which are incredibly small and have to be accurate

ceres

Yeah, biology is a kind of a messy business compared to chemistry. I love bio lectures, but I really miss chem lab. It was so much neater.

tofu

I always loved bio but there's a reason I decided to get a chem degree :P

Bliss 101

I don't think this is a math problem, but a problem with the experiment. The background absorbance depends on the solution the proteins are suspended in, and it sounds like you made the standard curve using pure BSA dilutions, whereas the unknown samples are cell lysates with all kinds of contaminants. I'd expect to see some kind of a protein purification step before any attempt to measure the concentration.

romanqwerty

Yeah, it sounds like the problem is that the make up of your standard curve is not equivalent to your samples. It sounds like you either need to purify your samples so that they are free of interfering species or find a way of replicating (or accounting for) those interferences in your standards.

ceres

Okay, see, that would make some sense. And for the purified step, sure, I can see doing something like this. But including every aliquot from every step into these caculations doesn't even seem worthwhile to me.

Like, if we're going to be obsessive about measurements, can't I just stick it in the damn nanodrop?

tofu

You don't necessarily need to purify them, you could have used a spiking technique to make your standard samples

mts

there is nothing wrong with using pure bsa for controls, you just need to make sure all standards are made up in whatever solute your samples are in.

i didn't really follow what you meant by getting a negative, can you try and explain the issue more clearly
you are not going to get absolute zero with your blank ever and i doubt you need to purify them, you are just trying to measure concentration.

Dibs

I would force your trendline to 0,0. Alternatively, don't use any of the dilutions that give you negative protein concentration values. I assume the dilutions are all from the same lysis sample, so just use the ones that work.

You're measuring total protein concentration in your samples - there's no need for a purification step. At most, you might centrifuge (10 min @ max speed on a tabletop) the samples down to isolate only the soluble fraction.

It's an undergrad lab, don't stress. You sound like you actually do have a decent grasp on what you're doing. It doesn't have to be perfect, the point of these labs is to make sure you understand how the Bradford assay works, how to use a standard curve, how to deal with dilution factors, etc. etc.

Sorry this was late.

Edit: nanodrop.. /drool

There are times you are going to need to use a standard curve to determine a value - the bradford is a super-widely used assay for protein concentration, making it a super easy way to familiarize students with standard curves.

mts

nanodrops work better for nucleotides, not so much for protein since you still need to do a standard curve.

i wouldn't worry about the trendline, the equation for the line will compensate for it when you calculate values.