Fig. 1.

The ith datapoint {DNA sequence ${\mathit{X}}_{\mathit{i}}$, positive reads ${\mathit{R}}_{\mathit{i}}{}^{+}$, negative reads ${\mathit{R}}_{\mathit{i}}{}^{-}$}, belonging to mode k is shown. The red (blue) box of width $w_k^{+}(w_k^{-})$ denotes the positive (negative) read window, which is modeled using Bernoulli probabilities ${\varphi }_{\mathit{k}}{}^{+}({\varphi }_{\mathit{k}}{}^{-})$. The 1 and 0 indicate the presence and absence of reads at those positions. The pink box is the DNA site at position Z_i, modeled using PWM parameters ${\varphi }_{\mathit{k}}$. The positive (negative) read window is at a fixed distance ${\tau }_k^{+}({\tau }_k^{-})$ from Z_i. Only ${\mathit{X}}_{\mathit{i}},\hspace{0.17em}{\mathit{R}}_{\mathit{i}}{}^{+}$ and ${\mathit{R}}_{\mathit{i}}{}^{-}$ are known; the rest are unknown and learned from the data