Contaminations in genomic sequences

Despite continued advances in whole genome sequencing techniques and the development of powerful assembly algorithms, newly sequenced genomes still often suffer from contaminations during the sequencing process. The most common sources of contamination are accessory DNAs deliberately attached to the DNA/RNA under investigation, including vectors, adapters, linkers and PCR primers. However, there are also unintended events, e.g. caused by transposon activity or simply impurities, leading to contaminated genomic sequences. These may then result in missassemblies of genomic sequences, meaningless analyses and potentially erroneous conclusions. However, no one knows to which extent publicly available genomes are contaminated. To encompass this unsatisfying situation we therefore plan to develop a comparative genomics approach to broadly identify contaminations in available genomic sequences. However there exist some tools those can find contaminations from adapters or from vectors alone. Here we present an approach based on machine learning to distinguish between contaminated and non-contaminated sequences instead of finding vector contaminations or adapter contaminations alone. As for now no such tool available, our approach would be foremost and showed promising results on different datasets.