Given the ATAC-seq data in open chromatin regions (OCRs) and the preferential DSB production in OCRs (nucleosome-free in particular https://academic.oup.com/nar/article/48/16/8993/5876288 ), it may be possible to predict (perhaps, the yield of, but not specific hotspot/fragile sequence for) some DSBs, albeit probably subject to chromatin conformational status at the time of exposure to a DSB-producing agent (e.g., ionizing radiation).

From my point of view, if you choose a very smart cell model for DSB or DNA repair, ATAC-seq (perhaps additional small RNA-seq) might be used for that prediction. For example, some biological process will generate many mismatched reads with this location accessible in the DSB or repair situation. In this backgroud, the reads produced by ATAC-seq not only represent how accessible the breaking sites but also incluedes the mutation information, which will be further used for DSBs prediction.

Thank you both Nobuyuki Hamada Wei Zhou for your insightful response. We have in vivo radiated livers harvested 24h after radiation stop, and the ATAC-Seq is performed in bulk tissue. We have surrendered our attempt to predict DNA damage in this material. However, I believe it would be possible if the experimental design is optimal (time point, cell type, scATCC-Seq, and adapted bioinformatic tools). I really liked your idea Wei Zhou regarding the mismatched reads, but I believe the background noise could make it difficult when bulk tissue is used.