I've been given a link to this 2019 study Effect of 900-, 1800-, and 2100-MHz radiofrequency radiation on DNA and oxidative stress in brain as evidence for some 5G nonsense conspiracy theory.
Even if the study has nothing to do with 5G, it makes a very important claim. Here's a copy of abstract:
Ubiquitous and ever increasing use of mobile phones led to the growing concern about the effects of radiofrequency radiation (RFR) emitted by cell phones on biological systems. The aim of this study is to explore whether long-term RFR exposure at different frequencies affects DNA damage and oxidant-antioxidant parameters in the blood and brain tissue of rats. 28 male Sprague Dawley rats were randomly divided into four equal groups (n = 7). They were identified as Group 1: sham-control, Group 2: 900 MHz, Group 3: 1800 MHz, and Group 4: 2100 MHz. Experimental groups of rats were exposed to RFR 2 h/day for 6 months. The sham-control group of rats was subjected to the same experimental condition but generator was turned off. Specific absorption rates (SARs) at brain with 1 g average were calculated as 0.0845 W/kg, 0.04563 W/kg, and 0.03957, at 900 MHz, 1800 MHz, and 2100 MHz, respectively. Additionally, malondialdehyde (MDA), 8-hydroxydeoxyguanosine (8-OHdG), total antioxidant status (TAS), and total oxidant status (TOS) analyses were conducted in the brain tissue samples. Results of the study showed that DNA damage and oxidative stress indicators were found higher in the RFR exposure groups than in the sham-control group. In conclusion, 900-, 1800-, and 2100-MHz RFR emitted from mobile phones may cause oxidative damage, induce increase in lipid peroxidation, and increase oxidative DNA damage formation in the frontal lobe of the rat brain tissues. Furthermore, 2100-MHz RFR may cause formation of DNA single-strand breaks.
I've read the full study, and unfortunately, while I'm an enthusiast, I'm not an expert, so I can't say more then "it looks good to me". Methodology looks good, 6 months of exposure is fine, obvious controls are there etc. An obvious limitation is that people are not mice, but mouse models are common. n=7 is also on the lower side, but it's probably enough for this kind of testing.
This seems a rather important conclusion, so either I'm missing something so it's not that significant, or it's a debunked, poor quality study or something similar.