Julien Mozziconacci (@Chromozz) header
Julien Mozziconacci (@Chromozz) profile picture

Professor in #Physics and #Biology @Le_Museum, interested in #chromosome architecture, #development #evolution and #deeplearning⌨️

Paris, France
Joined January 2017
Photo by Julien Mozziconacci (@Chromozz)Photo by Julien Mozziconacci (@Chromozz)Photo by Julien Mozziconacci (@Chromozz)Photo by Julien Mozziconacci (@Chromozz)Photo by Julien Mozziconacci (@Chromozz)Photo by Julien Mozziconacci (@Chromozz)Photo by Julien Mozziconacci (@Chromozz)Photo by Julien Mozziconacci (@Chromozz)Photo by Julien Mozziconacci (@Chromozz)Photo by Julien Mozziconacci (@Chromozz)
In our new preprint, we leverage #deeplearning to generate synthetic #DNA sequences that are able to form regular nucleosomal array with unatural repeat length in yeast. biorxiv.org/content/10.1101/…
1
7
1
44
9,870
Julien Mozziconacci retweeted
What happens to bacterial genomes when they're cloned into yeast? Do they transcribe, do they get silenced, do they get packaged away from the rest of the DNA? This cool study congrats to the Koszul lab team at Pasteur for this great work science.org/doi/10.1126/scie…
46
1
199
13,238
This fate can be predicted by a deep learning model trained on the host sequences only. This shows that deep genomics models can sometimes work even far away the training distribution. #deeplearning #Genomics
1
1
170
A speculative interpretation of our results would be that chromatin compartments appeared in eukaryotes following large DNA transfer events.
1
139
Julien Mozziconacci retweeted
[SAVE THE DATE] MLCB 2025 is happening Sep 10-11 at the NY Genome Center—NYC! Attend the premier conference at the intersection of ML & Bio, share your research and make lasting connections! Submission deadline: Jun 1 Details: mlcb.github.io Spread the word—please RT!
1
41
1
109
12,882
Julien Mozziconacci retweeted
New (and hotly anticipated - at least by me) preprint from my group describing a better way to partition training data for genomic-trained models to solve the long-neglected problem of homology-based data leakage. Thread from first author @muntakim_rafi 👇
0/ Essential reading for anyone training or using sequence-function models trained on genomic sequences! 🚨 In our new preprint, we explore the ways homology within genomes can cause leakage when training sequence-based models and ways to prevent it
2
11
33
5,390
Julien Mozziconacci retweeted
The Borzoi manuscript is now out in Nature Genetics: doi.org/10.1038/s41588-024-0… Borzoi predicts RNA-seq profiles in many tissues & cell types from DNA sequence as its only input. With it, we can score the impact of genetic variants on a number of gene-regulatory functions. 1/
2
42
1
172
14,695
Julien Mozziconacci retweeted
Curious about alternative splicing? Kick off the year with our latest review in @COSB_CRSB! authors.elsevier.com/sd/arti…
2
3
324
Julien Mozziconacci retweeted
Ledidi turns any genomics ML model into a controllable sequence designer by inverting the normal ML paradigm. Now, it is significantly faster, flexible, and more powerful than before. Available on GitHub and installable with `pip install ledidi`
2
22
2
121
11,269
Julien Mozziconacci retweeted
New preprint from our group on the genetic architecture of complex disease. Thread from 1st author @ChapelMadison 👇. Here, I give a some context of why I am particularly excited about this.
1 / Interested in how genetic variants influence complex traits? In our new preprint, we show that variant effects depend on the polygenic background in which they occur, and why it is critical to consider polygenic background when studying complex traits. biorxiv.org/content/10.1101/…
1
7
28
6,147
Julien Mozziconacci retweeted
How does chromatin linker length affect its emergent phase behavior? 🧬👩‍🔬 We explore this question via microscopy and MD simulations: Our new paper, in collaboration with the Rosen and Redding Labs, is out today!! 🔥 biorxiv.org/content/10.1101/…
3
15
57
9,020
Julien Mozziconacci retweeted
New preprint! "Gene-scale in vitro reconstitution reveals histone acetylation directly controls chromatin architecture", a tour de force project led by the amazing @FukaiTY, in collaboration with @KurumizakaLab (U Tokyo) and Umehara lab (RIKEN BDR). biorxiv.org/content/10.1101/…
3
52
4
181
34,666
Julien Mozziconacci retweeted
Interesting preprint on nucleosome spacing from cfDNA [Johnston et al., 2024] biorxiv.org/content/10.1101/…. The authors refer to NRL changes 10n -> 10n-5 bp. Earlier we showed that NRL decrease in cancer is more complex & context-specific [Jacob et al., 2024] link.springer.com/article/10…
7
37
4,670
Julien Mozziconacci retweeted
Chromatin architecture mapping by multiplex proximity tagging [Delamarre et al. 2024] biorxiv.org/content/10.1101/… ▶️Nucleosome maps in 3D with Proximity Copy Paste (PCP) ▶️Connectivity of nucleosomes in S. cerevisiae ▶️Chromatin is predominantly organized in regular nucleosome arrays
17
80
8,148
Julien Mozziconacci retweeted
Extensive binding of uncharacterized human transcription factors to genomic dark matter (Razavi et al., 2024) biorxiv.org/content/10.1101/… ▶️166 uncharacterized human TFs ▶️"Dark TFs", mainly bind closed chromatin enriched for transposable elements ▶️Some Dark TFs contain KRAB domain
1
51
1
253
20,607
Julien Mozziconacci retweeted
You still have a couple of weeks to submit an abstract for our next #LEGO meeting day in Paris! gt-lego.cnrs.fr/animation/le…
5
3
652
Julien Mozziconacci retweeted
For me one of the biggest unsolved questions in biology: how do different phenotypes arise and how is this related to genomic sequence, different cell types and gene regulation? We address this question in our BAT wing project. biorxiv.org/content/10.1101/…
1
26
1,027
Julien Mozziconacci retweeted
I've been looking forward to this release for a while: tangermeme v0.4.0 is out and asks "so... what did my model actually learn?" v0.4.0 moves past the individual components I've been building and shows how using them together can be very powerful. 1/ github.com/jmschrei/tangerme…
1
15
49
5,687
Julien Mozziconacci retweeted
Excited that CREME (Cis-Regulatory Element Model Explanations) is published in @NatureGenet! Work led by @ToneyanSh! tl;dr CREME interprets rules of gene regulation learned by genomic deep learning models using carefully designed in silico perturbations. rdcu.be/dT3kq
2
42
1
152
16,126