Did you do the sequencing? If so you can map the reads against the genome and search for islands or contigs with very different coverage

You can also redo the genome assembly trying to delimitate the reads to be used in each assembly based on the kmer frequency. You can do that with the velvet stats output file in R generating a graph:

>R --no-save --no-restore

>library(plotrix)

>data weighted.hist(data$short1_cov, data$lgth, breaks=0:50)

>dev.off()

Change the "breaks" parameter to change the range of your kmer coverage axis