Given two or more reads library sampled from their respective tissues (e.g. cancer or non-cancer), we have a method that can assign/classify the reads to their respective source genome. But without mapping to the reference genome. It also consider error rates in the reads and coverage.
This is inspired from the fact that there are no-identical genomes. Mapping them will lead to loss of information.
One possible application is to use it for finding rearrangement breakpoints reads. What are the other potential application of such methods?
Suggestion for related publications can be useful too.
To identify viral integration sites.
The landscape of viral expression and host gene fusion and adaptation in human cancer
This was not alignment free: "and further filtered of human content using Bowtie40. The prinseq-lite utility41 was used to remove low-complexity sequences (using a DUST threshold of 7) and short reads <45 nucleotides. Remaining reads were aligned to the RefSeq collection of viral genomes (n=3,590 excluding bacterial phages), downloaded on 19 Dec 2012."
I still don't understand how one can have "coverage" and classification to "source genome" withoug alignment.
Log in to answer this question.
how does this work without alignment?
Do you mean you have a method of removing contaminations or sequencing errors without mapping to a genome?
Are you looking forward at performing a de novo assembly ?