These tracks represent the FIRE peak calls, the inferred regulatory elements from Fiber-seq.
FIREs are MTase sensitive patches (MSPs) that are inferred to be regulatory elements on single chromatin fibers. To do this we used semi-supervised machine learning to identify MSPs that are likely to be regulatory elements using the Mokapot
framework and XGBoost
. Every individual FIRE element is associated with a precision value, which indicates the probability that the FIRE element is a true regulatory element. The precision of FIREs elements are estimated using Mokapot
and validation data not used in training. We train our model targeting FIRE elements with at least 90% precision, MSPs with less than 90% precision are considered to have average level of accessibility expected between two nucleosomes, and are referred to as linker regions.
Semi-superivized machine learning with Mokapot
requires a mixed-positive training set and a clean negative training set. To create mixed positive training data we selected MSPs that overlapped DNase hypersensitive sites (DHSs) and CTCF ChIP-seq peaks. And to create a clean negative training set we selected MSPs that did not overlap DHSs or CTCF ChIP-seq peaks.
Manuscript in preparation. Please refer to https://github.com/fiberseq/FIRE/tree/main/docs for more details.
Stergachis et al, Single-molecule regulatory architectures captured by chromatin fiber sequencing. Science 2022 DOI: 10.1126/science.aaz1646