Different Salt and/or Mg concentration?

To my eyes, attachments 2 and 3 both show a peak at 76.5; there is just better separation on attachment 3. I have never run a native gel, but my understanding is that unless the gel percentage is high enough, the gel is large enough, and the gel is run long enough it still might not be possible to separate minor differences in sequence. My guess is that you are amplifying the correct sequence, but there are polymorphisms within that sequence. If you have easy access to sequencing capabilities, you might want to sequence a couple of products with a peak at 74.5 and a couple with a peak at 76.5. (You might be able to use your PCR primers as sequencing primers to get a quick answer, although ideally sequencing primers are different from PCR primers.)

If you do find a polymorphism, you could keep your original PCR multiplex reaction or possibly you'd want to redesign those primers so that you amplify a region that does not include the polymorphism.

Dear Martin,

The assay was optimized with commercial mastermix (LC480 Sybr Green Master). The mastermix already optimzed by the company. And, only blind samples are different in these PCR reactions. =) 

Dear Deborah,

I used 8% native PAGE gel to screen the PCR product (Att4). It gives high resolution for 10-300bp amplicons separation and it can distinguish amplicon of 5-10 nt difference if I am not mistaken.

Lane L: 25bp marker; Lane 1-7: Sample 1-7; Lane 9: positive control; Lane N: NTC.

Thanks for the sharing =). I do agree with you that maybe there is a internal genetic variant which cause another Tm to jump up. Because multiple-copy gene is targeted, I was thinking 4 degree difference is huge. What come to my mind is that possible the few nucleotide differences can make such a big Tm value difference.

I think it is likely that the Tm variation is due to sequence variation in your amplicons. A change of A:T to G:C will typically give a Tm increase of 2 C, though it varies depending on the surrounding sequence ('stacking effects'). 

You will not see this on a gel, as gel electrophoresis separates fragments by size; sequence composition has very little effect on electrophoretic mobility. 

Do, however, remember that when curves overlap their peaks are drawn towards one another. This is a purely mathematical phenomenon. 

Hi Lee Lee,

There is web based application for melt curve analysis.

https://www.dna.utah.edu/umelt/um.php

Use it to make prediction and optimization of you assay. Please, bare in mind that this is just math.

Regards