Melting curve analysis: bioinformatic question (DAN) - (Sep/20/2006 )
I have recently begun using SYBR Green-based RQ-PCR and am having a bit of trouble interpreting my melting curve data. I am running two different reactions, one with a 200bp product and one with a 305bp product.
I have read that, for amplicons of <200bp, a single peak is expected, and that for larger amplicons a bimodal peak may be observed because of internal variations in GC content. So far, this matches my results. I have confirmed by agarose gel electrophoresis that there are no weird background products in either reaction.
However, I've been trying to go the extra mile and confirm independently that the PCR products should have the observed melting curve - esp. the biomodal one. I have been analyzing the product sequence using a DNA melting temperature calculation algorithm (http://bioweb.pasteur.fr/seqanal/interfaces/dan.html). My window size was 50 and shiftincrement 1.
Now my problem is analyzing the output from that! I'm having trouble interpreting the graphs. If anyone is familiar with this sort of thing, could you please tell me:
1) Is this a useful approach at all, or should I just relax?
2) Which, if either, of the two attached graphs would you say represents a DNA molecule that would yield a bimodal melting curve?
Thanks for the help, and for reading my long post!
Can you attach the original melting curves? I'm assuming - since you talk about peaks - that you're looking at the first derivative of fluorescent intensity [y-axis] versus temperature increments [x-axis]?
Have to say, I've never seen a bimodal peak and I've used SYBR green with amplicons up to 1 kb. Generally, above about 800bp the melt curve just gets too messy to yield any useful info but for under 500bp - provided the primers are designed properly and all - I've only ever gotten 1 peak.
Are your primer pairs designed so that at least one of them sits over a splice junction? If not then even a slight DNA contamination can cause problems but since you say there's no wierdness by agarose gel maybe this isn't the problem.
Hooly is right, shouldmlook at the original melting cureve which is fluorescent intensity [y-axis] versus temperature increments [x-axis].
By the way what real-time PCR machine are you using?
Thanks for the replies. Here's the melting curve, with the axes you describe. Having a look at it again perhaps "bimodal" was an exaggeration but I'd appreciate any comments.
The primers are for EGFP and the template is plasmid DNA, so different product sizes shouldn't be a problem.
The reactions were done on an SDS 7700. (In case you're wondering, the graph below is not a direct output of that instrument's software - due to computer issues too complicated and dull to get into here, I end up re-plotting the raw data in Prism.)
Yap, the bimodal peak is an exggeration