Chimeric sequences form during PCR amplification.
A chimeric template is created during one round, then amplified by subsequent rounds to produce chimeric amplicons that are difficult to distinguish from amplicons derived from a single biological sequence.
In each round, the number of amplicons of a given sequence approximately doubles (subject to biases due to sequence affinity and secondary structure), with the result that chimeric amplicons should almost always be less abundant than their parent sequences.
This observation is exploited by the de
novo mode of UCHIME. Given a set of
amplicon sequences, it is straightforward to find chimeric
alignments by using the reference database mode of UCHIME where the
same sequences are used for both database and query. The difficulty
is to determine which of the three sequences in the alignment is
the chimera and which is the parent. Amplicon abundances can be
used to discriminate: the parents should be at least 2x as abundant
as their chimeras. The ratio between the parent abundance and the
putative chimera abundance is called the abundance skew. The
minimum value required to classify a sequence as chimeric is set by
the ‑abskew parameter of uchime_denovo (default value