What I do when having many different samples is to add diluted samples to an inert 96 well plate (made of polypropylene, which does not bind proteins to a significant degree) in the same distribution that they will have in the ELISA plate (but in a volume slightly larger than the one I will use for ELISA). Once all the samples are there, it is vey easy and quick to transfer the desired volume (i.e. 50 microliters) from each well to the equivalent well of the ELISA plate with a multichannel pipette to avoid huge delays between the first and the last samples.

Agree with Josh, to add further, 

do you add a stop solution to stop the reaction? If so, do you add the stop solution in the same order as you added the substrate solution to the wells?

HI there

ELISAs are simple and straightforward, but there are some things one needs to be aware of.

In the final (colour development) phase, it's tempting to let the reaction run for "enough" time to get the colour in the readable range for the concentrations one is expecting. This is invariably a mistake, because it leads to samples and standards overrunning and moving out of the linear range.

So, as a first step to controlling your ELISA, ensure that you leave the final stage for at least 30 minutes in the dark and then add the H2SO4 (I'm assuming you're not using an Alk Phos linked ELISA here). If your controls - and there should be a standard curve - are overcooked, then this doesn't mean that your incubating the substrate for too long, it means that your concentrations of capture and detection antibodies are too high. If that's the case, go back and perform a checkerboard titration of these to elements against the same standard curve. This should almost certainly sort out your problem.

However, Rakesh raises an interesting point. If you are adding the substrate from (say) left to right across the plate and the stop solution from left to right, then your left hand wells will be cooking for longer than those on the right. However, you'll see this effect only if your colour reaction is proceeding too quickly (see above).

Good Luck!

G

In addition to the great comments already posted.

I have noticed when you make dilutions of you standard on the plate (mix up and down in the coated wells) then the blank one can be higher than some of the lower dilutions. Try to dilute before you add standard to the plates if you are not doing this.

On some type of plates I have noticed higher background in the surrounding wells which will give a high blank as this is most likely in a surrrounding well in your setup - long shot. Long time storrage of coated plates (not sealed correct) can give same phenomenon.

However, if some samples are 2.2 in OD and others are below blank, then I guess these are really low anyway and should maybe be noted as below LOD?

Your blank values are a bit high, you could try to reduce them by changing the blocking and diluting conditions, but as you also have very high samples, it is likely that your negative samples are really negative (no analyte or extremely low concentrations of it that cannot be detected). In practical terms, you could consider all these samples to have ''zero level''. Does it make biological sense that these samples do not contain the analyte? Otherwise, you would have to increase the sensitivity of your assay. 

First of all, thanks for this valuable advices

I already covered the plate from light after adding the substrate and waited for 15 min as the kit instruction then I added stop solution and measured the color intensity

I used multichannel pipette during the addition of substrate and stop solution and put them in the same order, but adding of samples themselves consumed about an hour because I use a fixed pipette to ensure all sample are taken.

Is this long period led to these puzzling results However, the negative cases is cases which put first on the plate

Also the sample diluted according to kit instruction 1:5 (10µ sample and 40 µ sample diluent)

Finally can I draw the standard. curve without subtract from blank and also determined the concentration of sample from the curve without subtraction from blank?

You can avoid substracting the blank, but have in mind that this is a trick to try to force the samples to be positive. The reality of your experiment is that the negative samples have no detectable analyte according to this assay (zero apparent concentration) so strictly you cannot quantify anything. Ans this is not so puzzling. Every assay has a useful range and a detection limit. If you expect that these samples have a much higher value, perhaps there is a real problem with the assay sensitivity. Is this the case?

What I do when having many different samples is to add diluted samples to an inert 96 well plate (made of polypropylene, which does not bind proteins to a significant degree) in the same distribution that they will have in the ELISA plate (but in a volume slightly larger than the one I will use for ELISA). Once all the samples are there, it is vey easy and quick to transfer the desired volume (i.e. 50 microliters) from each well to the equivalent well of the ELISA plate with a multichannel pipette to avoid huge delays between the first and the last samples.

Another big question is, what else is in the samples besides the specific target antigen. If it's very important that this should be quantitative, you should not only have an ELISA buffer blank control, you should also have a good negative control that consists of the same matrix that your samples are in. There are some assays where I have seen matrix effects even at what one would have thought were fairly high sample dilutions and it was like your scenario where the samples are coming out unexpectedly low, because something in the matrix was reducing target antigen binding.

For example if the samples are from cell culture, include a medium control. If the samples are some other preparation e.g. experimental vaccine, the blank should be a mock vaccine formulation including all salts, adjuvants, stabilizers etc. minus the antigen.

@Gertrudis Rojas thanks for our precisely advice

I diluted the sample 5 times .I wonder, if I did not dilute them the O.D of my sample will be higher than blank O.D

can I multiply the O.D of sample by 5 instead of multiplying the concentration of samples (because after subtract the answer will be -ve and multiplying will not be useful )

No, multiplying the absorbance is not a valid approach. You dont know what would be the abosorbance without diluting, it doesnt have to be 5-fold higher. I think you should accept that this assay was not useful to determine the concentration of certain samples. It can be due to a very low analyte concentration, to  matrix effects as explained above or to low sensitivity of the assay.

Has to be non-specific binding of either primary or seconday antibody.  Two suggestions:

Soak everything and run the assay in 1% BSA

Send the binding pair off to a biacore facility, either a colleague or commercial outfit and have them immobilize to a CM5 chip and test against analyte.

Forget the blank value. You get this from a well (either coated or uncoated), where nothing else even no enzyme was insite. So you can get undefined reactions which can not be controlled.

You need th have the 0-value, everything was in the well but no analyte. The whole matrix, the buffer(s), the conjugate and the subtrate as well as the stopping solution. 

For the measurement, this value (zero concentration of the analyte) should be lower than 0.1, the max value should be between 1.5 and 2.0 (just for optical reasons). This will result in a P(Positive)/N(Negative)-ratio higher than 15. 

Can we include Negative samples in ELISA after zero correction for analysis or in stat or for mean calculation by considering that sample reading as 0. Any reference?

I would consider negative concentrations as zero.