1. You should compare serum and biopsy. Then determine which sample is more relevant for your situation.
2. Determine the sequence you want to target. Design primers using MethPrimer - http://www.urogene.o.../methprimer.cgi
. Input DNA, the software will bisulite convert in silico and design the primers. Try to keep the amplicon size around 400-600bp. Bisulfite convert your DNA with a kit or make the reagents. PCR amplify your bisulfite converted DNA with primers designed by MethPrimer. Run the product on the gel and gel purify the band. Clone this into a vector (i.e. TA Cloning). Picking 10 individual colonies (i.e. 1 clone would be equivalent to 1 allele) and sequence the insert from the isolated plasmid (traditional sanger sequencing). Your sequencing core may have software to help with the peak calling. You can then compare the 10 sequences from one sample using CpGViewer - http://dna.leeds.ac.uk/cpgviewer/, to get an overview of the methylation status. 10 sequences per sample would allow you to determine the methylation status of 10 alleles.
3. The taq does not matter. If you are going to perform TA cloning then a non proofreading taq is needed to add the A overhangs. Otherwise standard taq works fine. Hot start may help but is not necessary.
Methylated and unmethylated primers are for methylation specific PCR. F primer overlaps a CpG site at 3' end. Two F primers would be designed - methylated will have a CG and unmethylated a TG. You would run two PCRS, one with methylated primer and one with unmethylated primer. If CpG is methylated then only the methylated primer would anneal and produce a PCR product, while he unmethylated primer would not anneal and no product. Only allows you to interrogate a single CpG. Bisulfite sequencing is much more information but more work.
Hope that helps!