Remember, epigenetics also reflects the exposure of the indvidual to the environment, to expect low "residual confounding" given the nature of this is just not realistic.

My advise if you review an #EWAS paper is to judge the plausibility of the results, the biology and the covariates used. And of course wether it is a big meta-analysis (a sample size of 1000 is enough to find robust findings) and/or replicate their results. We should stop to ask for pointless work

Nevertheless, we highlight that our findings provide a valuable foundation for future research aimed at understanding the #epigenetic mechanisms underlying physical activity and its potential impact on health.

We identified 122 CpG sites associated with moderate activity and 2 with sedentary behaviour. However, we did not find evidence of replication in DNA methylation data of adolescents from ALSPAC. Hence, we need a cohort with similar sample size to further conclude.

The paper was published in the prestigious journal #Diabetes, published by the American Diabetes Association #ADA.

We managed two replicate 5/6 hits, of which 3 were in TXNIP, a gene heavily triggered by glucose. The CpG identified have been previously related with type 2 diabetes, diabetic kidney disease and fasting insulin. Overall this research contributes to elucidate the epigenetic component of IR.

Forgot to add all the right emojis, but of course this work needed a strong competence of biostatistics and bioinformatics to be carried out!! 🖥️🧬

These studies may show the importance of having a diverse sample in EWAS studies. Of course, we should be careful in designing our model, as we should handle ancestry differences correctly. But, if handled right, it seems that is a good approach!!

In line with this hypothesis, we have a previous article that consisted on an EWAS of insulin resistance during pregnancy published in Diabetes. Here we show that 80% of the cross-ancestry CpG 🧬 sites were replicated, while the "ancestry specific" were not

Our hypothesis is that, if we find a signal has the same effect across ancestries, despite the underlying difference in methylation and genetics due to ancestry, this CpG site would be a true finding indeed.

So this is very good news as we know that in general there is reproducibility crisis in scientific papers. We believe that the high reproducibility of the findings is due to our cross-ancestry approach. For discovery we used the STORK-G cohort, which consist in  European and South Asian.

Of the identified 39 CpG sites associated with BMI in maternal blood during pregnancy, we managed to replicate 13 CpG sites in the Norwegian MoBa cohort. This is cool, as someone in the field may now, it is difficult to replicate EWAS findings.

Best t-shirt eve

These studies may show the importance of having a diverse sample in EWAS studies. Of course, we should be careful in designing our model, as we should handle ancestry differences correctly. But, if handled right, it seems that is a good approach!!

In line with this hypothesis, we have a previous article that consisted on an EWAS of insulin resistance during pregnancy published in Diabetes. Here we show that 80% of the cross-ancestry CpG sites were replicated, while the "ancestry specific" were not.

Our hypothesis is that, if we find a signal has the same effect across ancestries, despite the underlying difference in methylation and genetics due to ancestry, this CpG site would be a true finding indeed.

So this is very good news as we know that in general there is reproducibility crisis in scientific papers. We believe that the high reproducibility of the findings is due to our cross-ancestry approach. For discovery we used the STORK-G cohort, which consist in European and South Asian.