EMBL Seminars

Abstract
Analysing multi-omics datasets is cumbersome and requires significant bioinformatic skills. Illumina Connected Multiomics™ (ICM) is designed to simplify the analysis and generate valuable insights from multiple omics datasets. This hands‑on learning session provides researchers with guided training on how to use ICM for real‑world analysis workflows. Participants will practice navigating the platform, running analyses, and interpreting results across -omics modalities.

Illumina Connected Multiomics (ICM) is a secure, cloud-based analysis and data management platform designed to simplify and accelerate multi‑omics workflows. ICM enables researchers to seamlessly integrate, visualize, and analyze sequencing data across genomic, transcriptomic, epigenomic, and proteomic applications. With intuitive interfaces, scalable compute, and built‑in best‑practice pipelines, ICM empowers users to generate insights faster—no advanced bioinformatics expertise required.

Agenda

• Multi-omics introduction & roadmap – 20 min
• ICM introduction – 15 min
• Getting started: workflow navigation, platform login, data import - 25 min
• Coffee break & troubleshooting – 10 min
• MOx analysis (bulk RNA + miRNA) – 45-60 min
• Demo single-cell, spatial, epigenetic, proteomics data – 45 min
• Q&A and best‑practice recommendations – 10 min

Capacity: 20 attendees

(We only have 20 seats available. You will be notified per email about status of your registration.)

Please register via this page. 

Abstract

Alzheimer’s disease (AD) is a highly prevalent neurodegenerative disease that exclusively affects

elderly people. Here, we used direct conversion of primarily sporadic AD patient fibroblasts into

induced neurons (iNs) to generate an age-equivalent neuronal model. Patient-derived iNs exhibit

strong AD-specific transcriptome neuronal signatures characterized by down-regulation of mature

functional and morphological properties and up-regulation of immature neuronal and neural stem

cell-associated pathways. Mapping AD and control iNs to longitudinal transcriptome data from

maturing human neurons demonstrated that AD iNs are fully converted into iNs, but reflect a dedifferentiated

neuronal identity. Epigenetic landscape profiling revealed an aberrant cellular

program underlying their immature neuronal state, which shares similarities with malignant

transformation and age-dependent epigenetic erosion. To probe for the involvement of aging, we

generated iPSC neurons from the small cohort, which, indeed, showed non-significant diseaserelated

transcriptome signatures. This is consistent with epigenetic aging clock and brain

oncogenesis mapping, which indicated that unlike iPSC neurons, iNs more closely reflect adult

and old brain stages, rendering them a valuable tool for studying adult-specific, age-related

neurodegeneration. In this model, AD-related neuronal changes appear less as a mere

accumulation of damaging events, but rather an age-dependent cellular program that impairs

neuronal identity.
 

This seminar will take place 

Abstract
In this webinar James Malone, co-founder and CTO of Luvida, an AI-driven platform improving clinical trial design, will share insights from his journey from EMBL-EBI to entrepreneurship. Drawing on his experience from early bioinformatics work at EMBL-EBI to applying AI in clinical trial design, James will share how advances in data and technology have reshaped what’s possible and why real-world impact still lags behind. 

A more detailed description, can be found in the registration link below.

The webinar will be live-streamed from EMBL-EBI, and participants working on the Genome Campus are welcome to attend in-person in the Dendron Seminar room.

Registration - and option for in-person attendance
Please register for the webinar via zoom, both for in-person attendance and online participation.

https://embl-org.zoom.us/webinar/register/WN_7kAz3LykT5OIuyjXpjLPlA#/registration

Please note that the talk will  be recorded.

Two-dimensional template matching (2DTM) has proven effective for in situ identification of large soluble proteins in cryo-EM micrographs of cellular slices. Detecting membrane proteins, however, remains challenging because their lower molecular weight leads to weaker signal-to-noise ratios in crowded cellular backgrounds, while the high contrast of lipid bilayers can generate spurious low-resolution matches and high false-positive rates.

In this talk, I will present a workflow that improves true-positive membrane protein detection while reducing membrane-artifact false positives. Rather than applying hard thresholds to matching scores, we developed a two-stage, multi-branch neural network for particle classification. In the first stage, a convolutional neural network processes 2DTM-derived feature maps to generate an initial classification score and a compact embedding summarizing the 2DTM evidence. In the second stage, candidates passing the first stage are further evaluated using particle-centered and larger-context image patches, whose representations are integrated with the 2DTM embedding for final classification. This framework improves recognition of membrane proteins across a wider range of views, including both side and top views.

The network is trained entirely on a physics-based simulated dataset and generalizes to experimental data of proteins embedded in liposomes. In benchmark tests, the degree of template bias decreased from ~60% to ~30%, representing a substantial reduction in false-positive influence. Together, these results show that combining 2DTM with deep learning can enable more reliable membrane protein detection and improve downstream structural analysis.

Abstract
Our research focuses on the role of RAF kinases in transmitting signals within the RAS-RAF-MEK-ERK (RAS/ERK) cascade. Hyperactivation of the RAS/ERK pathway caused by activating mutations in RAS and RAF is a major driver of tumor formation in ~30% of all cancers1. Improving our understanding on the regulation of the RAS/ERK pathway is of paramount importance for the development of next-generation therapeutics.  

The RAF family comprises three catalytically active isoforms (ARAF, BRAF, CRAF) and two pseudokinase isoforms (KSR1, KSR2), which adopts dimers conformation to activate RAF catalytic outputs2. Our goal is to investigate the molecular mechanisms governing the dimerisation of KSRs with RAFs. Our investigation uncovered an allosteric mechanism driving KSR1:BRAF heterodimerization and BRAF activation, which depends on MEK binding to KSR and on the interactions between the N-Terminal domains of KSRs and BRAF3. Additionally, we revealed how the scaffold proteins CNK and HYP potentiate this KSR-dependent mechanism through the formation of an unexpected CNK:HYP:MEK:KSR quaternary complex4.

Employing integrative structural biology techniques combining NMR, X-Ray crystallography and cryo-EM alongside biochemical analyses, we aim to deepen our understanding of the pivotal role of the pseudokinase KSR in regulating the RAF kinases, providing novel insights into the multi-layered regulation of the oncogenic RAS/ERK signaling pathway.

1Lavoie H et.al., Nat Rev Mol Cell Biol (2015) - 10.1038/nrm3979 
2Rajakulendran T et.al., Nature (2009) -10.1038/nature08314
3Lavoie H et.al., Nature. (2018) -10.1038/nature25478
4Maisonneuve P et.al., Nat Struct & Mol Biol (2024) -10.1038/s41594-024-01233-6

Connection details
Zoom*: https://embl-org.zoom.us/j/97170620085?pwd=LklFaC4gFmlJHvw5tMSbT2vqaKsaNV.1   
Webinar ID: 971 7062 0085
Passcode: 717935

Please note that the talk will be recorded.
*For the FAQ section, as a zoom participant, please use either the chat function (the host will read out your question) or the “raise your hand” function and turn on your microphone.

Abstract
To be announced

Connection details
Zoom*: https://embl-org.zoom.us/j/97170620085?pwd=LklFaC4gFmlJHvw5tMSbT2vqaKsaNV.1  
Webinar ID: 971 7062 0085
Passcode: 717935

Please note that the talk will be recorded.
*For the FAQ section, as a zoom participant, please use either the chat function (the host will read out your question) or the “raise your hand” function and turn on your microphone.

Abstract

Epigenetic components regulate many biological phenomena during development and normal physiology. When dysregulated, epigenetic components can also accompany or drive diseases. One main class of epigenetic components are Polycomb group proteins. Originally, Polycomb proteins were shown to silence gene expression. We found that this function involves the regulation of 3D chromosome folding and we found that Polycomb components can induce the formation of long-distance interactions or chromatin loops that may play instructive roles in gene regulation as well as serve as scaffolding elements that contribute to enhancer-promoter specificity. Perturbation of Polycomb components is involved in human cancer and leads to tumorigenesis in flies. Surprisingly, even upon a transient depletion followed by restoration of the full Polycomb compendium, epithelial cells lose their normal differentiated fate, continue proliferating and establish aggressive tumors, demonstrating that cancer can have a fully epigenetic origin. Similarly, transient perturbation of histone acetylation in mouse ES cells and gastruloids shows that they can record chromatin changes and that this results in cellular memory of the perturbation states. The implication of these data will be discussed.