We present a kmer-based heuristic approach for partitioning sequencing fragments based on 
a set of reference genomes to address potential contamination from multiple sources. The 
method involves constructing a k-mer index for reference genomes, processing each fragment 
by extracting all possible k-mers and searching for matches in the k-mer index. Then, we 
calculate the proportion of matching k-mers and the coverage for each genome, representing 
the percentage of k-mers from the reference genome corresponding to the k-mers in the 
fragment.

To classify each fragment into one of three categories (unmapped, unique, or ambiguous), 
we utilize two predefined thresholds: proportion and coverage. In this study, we employed 
default thresholds of 0.1 

for coverage and 2 for proportion. The coverage threshold of 0.1 was selected to balance 
sensitivity and specificity when evaluating the reference genome's coverage by the 
fragment's k-mers. Conversely, the proportion threshold of 2 was chosen to ensure that a 
fragment's k-mers found in one genome is at least twice the proportion found in any other 
genome, providing a conservative approach for classifying fragments as unique or 
ambiguous.

Fragments with no k-mers found in any genome are classified as 'unmapped,' whereas 
fragments that surpass both thresholds for a single genome are classified as 'unique' to 
that genome. fragments that match multiple genomes but fail to pass both thresholds for 
any specific genome are classified as 'ambiguous.' We partition the fragments into groups
corresponding to each category for every reference genome, allowing ambiguously classified
fragments to be duplicated across multiple genomes.