Figure 2.

(a) Extracted cfDNA is divided into two molecular pools, each containing a fraction of the tumor-derived cfDNA (red fragments). Each molecular pool undergoes a targeted amplification reaction, resulting in the introduction of both stochastic (red) and recurrent (pink) artifacts. Post amplification, each molecular amplification pool (MAP) is indexed and sequenced. (b) Variant data from the two molecular pools is statistically compared to eliminate stochastic errors and referenced to a large number of reference molecular pools to eliminate recurrent errors.