The difference between autopolyploidy and allopolyploidy lies in the evolutionary distances between the ancestral genomes. For autopolyploids, the two copies of the genomes came from the same species or even the same individual. So the subgenomes are expected to be very similar. The only way to distinguish autopolyploidy and regular diploidy would probably be querying the allele frequencies of variants detected. If substantial deviation from 0.5 (heterozygosity in regular diploid organisms) is observed, it could indicate possible autopolyploidy. It should be noted the pattern has to be global, because local copy number variations generate similar deviations from heterozygosity, too.

For allopolyploids, I'd apply the same principle of allele frequency deviation, but expect much more differences between ancestral genomes. It becomes really tricky, because fixed differences between two ancestral genomes now look identical with regular heterozygous variants in a diploid. For selfing species, an elevated level of heterozygosity may indicate possible allopolyploidy, while for outcrossing species, you may need to estimate the overall variation level, an increasing level of overall variations, combined with deviation from 0.5 allele frequency would suggest possible allopolyploidy. The good news is that divergence outside genic regions increases very fast following speciation. Combining two genomes separated a couple millions years ago to form an allopolyploid would give you a lot of highly divergent regions outside genes. You may be able to find signatures of such regions from your data.