Do DNA Break Sites Systematically Weaken Nearby TAD Boundaries? Comparing Insulation Scores Across 79 Damage Sites

Context

What the system looked like before.

Alongside the CTCF poster work, I worked with Heng Li on his research into how the 3D human genome structure responds to DNA damage induced by radiation, to understand how these disruptions affect diseases like cancer. Heng's research compares two DNA strands, mOHT (undamaged) and pOHT (damaged) — the damage created at specific double-strand break sites, or DSB sites, by the AsiSI restriction enzyme. From there, the research studies the strength of the TAD (topologically associating domain) boundaries around the DSB sites, looking at how CTCF and the cohesin complex work together to regulate gene expression at those sites.

Question

Does the strength of TAD boundaries around a DNA double-strand break systematically change compared to undamaged DNA — and if so, in which direction?

Approach & rationale

Working alongside Heng, I worked on both data analysis and wet-lab support. For the data analysis, I compared the two DNA strands, investigating their locations and comparing characteristics like Insulation Score to help understand the effects of the damage, using Excel and R Studio to build box plots. In the wet lab, I worked on cell culture and gel electrophoresis — growing GM12878 cells and learning qPCR — to support a starvation arm of Heng's research: starving cells before DNA damage to see how they react differently than non-starved cells.

Methodology & iteration

Including the paths that did not hold.

Pipeline: identify the ~80 AsiSI-cut DSB sites → define a matched genomic window in the damaged (pOHT) and undamaged (mOHT) condition for each site → compute an Insulation Score for each window from Hi-C data → take the difference (pOHT − mOHT) as a measure of how much the TAD boundary near that break site changed under damage.

Results

Numbers first, not buried in prose.

79 DSB sites were mapped, yielding 158 computed Insulation Score differences (damaged minus undamaged) across flanking TAD-boundary windows. 114 of the 158 (72%) are negative — insulation score reads lower, the boundary weaker, in the damaged condition at most of these windows — versus 44 positive. Mean difference −0.065, median −0.051.

This is a direct summary of the spreadsheet, not a statistical test — no significance test has been run against these differences yet, so it points toward a direction rather than confirming one.

Significance

Where this went.

Proposed for presentation at the EUReCa undergraduate research exhibit as part of the AURA award. Whether this specific analysis was presented separately from the CTCF poster is still being confirmed.

Reflection

This analysis ran alongside the CTCF/glucose-starvation work above, under the same AURA-funded collaboration with Heng Li. A fuller reflection on this piece specifically is still being written.