Gherman Novakovsky (слава Україні! 🇺🇦) Profile
Gherman Novakovsky (слава Україні! 🇺🇦)

@NovakovskyG

Followers
277
Following
3K
Media
23
Statuses
749

PhD, Illumina AI lab; interested in Deep Learning and genome regulation; also drawing, martial arts, guitar, and death metal! (he/him)

Joined January 2018
Don't wanna be here? Send us removal request.
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
Excited to share my first contribution here at Illumina! We developed PromoterAI, a deep neural network that accurately identifies non-coding promoter variants that disrupt gene expression.🧵 (1/)
Tweet media one
2
35
115
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
23 days
RT @KuanHaoChao: Excited to introduce LiftOn – an open-source tool for accurate liftover of genome annotations (GFF) across assemblies. 🚀….
0
18
0
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
RT @RNA_Life: Congratulations Gherman! 🖥️🧬🥳 A tour-de-force of AI/ML on predicting promoter variant effects in humans. 🔗: .
0
2
0
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
Huge thanks to the amazing Illumina team—this was an incredible learning experience! I'm excited to keep pushing forward as we develop models to tackle gene expression and non-coding variant interpretation. (16/).
0
0
2
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
A complementary thread from my colleague Kishore Jaganathan @kjaganatha (15/).
@kjaganatha
Kishore Jaganathan
1 month
We're thrilled to introduce PromoterAI — a tool for accurately identifying promoter variants that impact gene expression. 🧵 (1/)
Tweet media one
1
1
3
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
Want to learn more about PromoterAI?.📄 Read the paper: 💻 Explore the code & precomputed scores: (14/).
1
1
4
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
We followed up by testing promoter variants in Mendelian genes using MPRA. Surprisingly, PromoterAI was more effective than MPRA at prioritizing variants linked to patient phenotypes, highlighting limitations of MPRA for rare disease interpretation. (13/)
Tweet media one
1
0
1
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
While we noticed that the use of additional species such as mouse does not lead to substantial improvement of variant effect prediction, it does help with ensembling. Thus, the final model is an ensemble of two: trained on human only and trained on mouse+human together. (12/).
1
0
1
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
In the Genomics England rare disease cohort, functional promoter variants predicted by PromoterAI were enriched in phenotype-matched Mendelian genes. These variants accounted for an estimated 6% of the rare disease genetic burden. (11/)
Tweet media one
1
0
0
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
In the @uk_biobank cohort, PromoterAI's predicted promoter variant effects correlated strongly with measured protein levels and quantitative traits, suggesting that promoter variants contribute meaningfully to phenotypic variation in the general population. (10/)
Tweet media one
1
0
1
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
PromoterAI's embeddings split promoters into three distinct classes: P1 (~9K genes, ubiquitously active), P2 (~3K genes, bivalent chromatin), E (~6K genes, enhancer-like). The E class, enriched for TATA boxes, may reflect enhancers co-opted as promoters. (9/)
Tweet media one
1
0
2
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
Fine-tuning improved PromoterAI’s ability to predict the direction of motif effects — a known issue of multitask models. The model often recognized motifs before fine-tuning, but got the direction wrong. After fine-tuning, its predictions aligned better with the data. (8/)
Tweet media one
1
0
1
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
We used our list of gene expression outliers to explore their effect on transcription factor binding sites. Our results show that it is easier for new variants to cause outlier gene expression by disrupting existing regulatory components rather than creating new ones. (7/)
Tweet media one
1
0
2
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
We also attempted to fine-tune Enformer and Borzoi on our promoter variant set. While performance improved, both models lagged behind PromoterAI. Notably, PromoterAI outperformed Enformer and was similar to Borzoi before fine-tuning. (6/)
Tweet media one
1
0
2
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
When it comes to predicting expression effects of promoter variants, PromoterAI achieved best performance across benchmarks spanning RNA, proteins, QTLs, and MPRA. (5/)
Tweet media one
1
0
2
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
The second step was to fine-tune the model using a carefully curated list of rare promoter variants linked to aberrant gene expression. The fine-tuning was done using a twin-network setup to ensure the generalization across unseen genes and datasets. (4/)
Tweet media one
1
0
3
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
First, we pre-trained PromoterAI to predict histone marks, TF binding, DNA accessibility, and CAGE signal from a genomic sequence. The key difference with models like Enformer and Borzoi is that we predict at a single base-pair resolution and use only TSS-centered regions. (3/)
Tweet media one
1
0
3
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
PromoterAI is built from transformer-inspired blocks called metaformers — but instead of attention, we use depthwise convolutions, making it a fully convolutional model. We believe that CNN-based methods are not surpassed yet and remain a great choice for genomics tasks. (2/)
Tweet media one
2
13
49
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
1 month
RT @illumina: Today, we unveiled PromoterAI, a groundbreaking algorithm that, for the first time at scale, accurately deciphers pathogenic….
0
11
0
@NovakovskyG
Gherman Novakovsky (слава Україні! 🇺🇦)
2 months
RT @RNA_Life: Thank you for this amazing opportunity, and congratulations to all the new Azrieli Scholars!. I'm excited to contribute to su….
0
2
0