I would usually write a summary but I will defer to the text this time and just express thanks for the interactions and experiences this work generated. Its been a gateway into a lot of fun science that we are still unpacking...

It also required an amazing set of deep collaborations: particularly @daniel-decarvalho.bsky.social
(whose work motivated a lot of our thinking) and John Lacava who went all in for this work with their great labs and colleagues.

Trying to answer these questions required a theoretical framework at the intersection of physics, machine learning and evolutionary dynamics that our lab most enjoys - led by Petr Šulc, long-time collaborators at Ecole Normale in Paris, and our amazing lab at @mskcancercenter.bsky.social

* How, exactly, do we quantify viral mimicry: the expression of features associated with "nonself "viruses by our own "self" cells?
* Why do we see mimicry in so many disease settings such as cancer?
* How broad a phenomena is this?
* How do we think about selection in this evolutionary context?

Most importantly our work could not have happened without the support of the National Cancer Institute. The ability to receive NIH funding and interact with our colleagues there was critical to the growth of this work and our lab.

Understanding LINE-1's role in cancer is therefore a fundamental evolutionary problem and offers a potential therapeutic index for immune and other strategies that target its mechanism (such as the RT pathway see below) 👇👇👇

www.nature.com/articles/s41...

We found other genes associated with LINE-1 beyond p53.

We also found activity in Li-Fraumeni Syndrome, a pre-disposition syndrome associated with p53 mutations. Finally, we deposited our estimates of germline activity on dbGAP, one of the largest datasets on the topic:

ncbi.nlm.nih.gov/gap/

The main predictor of LINE-1 activity is if p53, the most mutated gene in cancer, is mutated. This is a deep evolutionary relationship. We mathematically modeled the dependence of LINE-1 action on p53 status, finding a mixture of transcriptional and genomic regulation:

We were able to rank tumor types by their "RT burden" - the degree to which new LINE-1 insertions happen in the self genome & also capture locus specific transcriptional heterogeneity:

We were able to quantify lots of interesting events, such as the rate LINE-1 inserts inverted copies, important for the generation of #viralmimicry through creation of "non-self" double-stranded RNA in the self genome.

We recently explored this phenomena in cancer below:

cell.com/immunity/pdf/S

We quantified the landscape of LINE-1 activity in the now ~5000 whole genomes and transcriptome datasets in the Cancer Genome Atlas. We built our own caller TotalReCall to measure how often LINE-1 copies itself in cancer. TotalReCall is publicly available:

github.com/Rome-Tx/tota...