Fig. 2

A proof-of-concept example of the proposed method for normalizing degradation patterns. a Expected or theoretical read coverage curves of one gene from four samples of identical shape (solid lines, without degradation), two of which (S1 and S2) are subject to degradation according to a rate indicated by the dashed lines in the 5′ end. b A realization of the four coverage curves without degradation randomly simulated according to the expected curves in a. These curves are regarded as the latent and unobserved data. c Observed coverage curves after imposing random degradation to S1 and S2 (S3 and S4 stay intact). d Estimates of the non-degraded latent curves from the proposed algorithm solely based on the observed coverage curves in c. e A sample-by-sample comparison between the observed and estimated latent coverage curves