Seminars, events & talks

Friday, 23th July, 2021, 13:00

Identifying longitudinal psychiatric signatures in HD using dynamic time warping

While Huntington's disease (HD) is diagnosed by the onset of motor symptoms, psychiatric disturbances bear a greater burden on daily functioning. Such psychiatric features can present up to fifteen years prior to formal disease diagnosis, although there is great inter-individual heterogeneity in symptom expression and evolution. As such, the present study strives to discern longitudinal psychiatric signatures that inform patterns of HD progression. Forty-seven HD gene-expansion carriers (23 premanifest, 24 manifest) underwent evaluation for depression, irritability, apathy, and executive dysfunction using the short-Problem Behavior Assessment (PBA-s) for a maximum total of six longitudinal visits. Unsupervised clustering of weighted PBA-s scores was performed with an adapted Disease Trajectories analysis software based on dynamic time warping (DTW), a technique that allows the non-linear alignment of sequences that may vary in overall lengths or speed, but can conceal the same temporal characteristics. This proof-of-concept allowed the identification of clusters with shared patterns of psychiatric evolution. Next, selected clusters were assessed for group differences in diagnostic status and severity of psychiatric features. This seminar will serve as a rehearsal of my Master's defense, which I completed under the supervision of Dr. Alexia Giannoula and Dr. Laura Furlong. In this way, I hope to stimulate a discussion regarding the analytical pipeline and its applicability and interpretability in the clinical field. Link:

Speaker: Audrey De Paepe, Master of Multidisciplinary Research (BIST), Cognition & Brain Plasticity Unit (UB) - Research Programme on Biomedical Informatics (GRIB, IMIM/UPF)

Room Online

Tuesday, 22th June, 2021, 10:00

Bioinformatics, combining facility and research

OpenGRIB Seminar. Bioinformatics is a common need in most research labs. Therefore it is important to have access to a core facility that is able to implement the standard methodology to extract knowledge from biological data, including omics data. In addition, new methods are continuously developed in the field to meet increasing data analysis demands. In this context, we will explain our experience as a service oriented facility and will also see some of the recently funded projects we are involved in, coming out of fields as disparate as COVID-19, bladder cancer or major depression disorder. Link:

Speaker: Lara Nonell, PhD BiCoH (Human Computational Biology), MARGenomics, IMIM Scientific and Technical Services

Room Online

Thursday, 20th May, 2021, 12:00

Identifying genomic variants acting on brain specializations

PRBB Computational Genomics Seminars, Chair: Mar Albà (Head of Evolutionary Genomics Group) Humans possess a nervous system which confers very distinct cognitive abilities and very distinct cognitive disorders. The study of brain development is critical for the understanding of the evolution of these distinct features. To discern the genetic causes in evolution and disease influencing human-specific phenotypes one needs to identify the relevant variants affecting relevant genes among thousands of other variants predicted to be neutral. Among the approximately 35 million single nucleotide polymorphisms (SNPs), 5 million insertions or deletions (indels), and 90 megabases of structural variants where the human and chimpanzee genomes differ are countless variants associated with development, function or disease. However, identifying these rare variants from among the thousands of variants expected to be neutral is a herculean task, a truly "needle-in-a-haystack" scenario. Many of us have reasoned that identifying evolutionarily relevant genetic variants, as well as those implicated in disease or function, can be guided by the analysis of species differences in intermediate molecular phenotypes (e.g., transcriptomic and epigenomic signatures), which are most likely the primary effects of genomic variation. Going one step further in the strategy to identify evolutionarily relevant variants, one can interrogate divergence in the segments of the predicted regulatory elements that are directly functionally affecting gene expression: the transcription factor binding sites (TFBS). The study of the brain spatiotemporal convergence of risk for multiple neuropsychiatric traits has pointed to a reduced number of transcription factors with critical involvement in normal neurodevelopment. We aim to dissect the temporal dynamics of genome-wide transcription factor binding site occupancy for a selection of risk convergence transcription factors (TF) at different stages of neurodevelopment and with an evolutionary perspective.  But even when the connection has strong probability of causality, one still needs to demonstrate the mechanisms underlying the phenotypic change, which is an effort rarely pursued by evolutionary biologists. This project aims to deal with these two problems by i) reduce the search space for relevant variants into a tractable list, and ii) test their functional effects in a system proximal to human fetal brain development consisting in iPSC-derived neural cultures. Zoom webinar:  / Password: 788137

Speaker: Gabriel Santpere, Head of the Neurogenomics Group - Research Programme on Biomedical Informatics (GRIB, IMIM/UPF)

Room Online

Tuesday, 27th April, 2021, 10:00

Taking full advantage of Real World Data through Biomedical Informatics

OpenGRIB Seminar. Real World Data (RWD) in Medicine is data derived from different sources such as structured data from electronic medical records, clinical annotations free text, patient-reported outcomes or even social media. The use of Biomedical Informatics tools is critical to make the most of extracting, analyzing and understanding health information from RWD. In this session, we will give an overview of some applications of Biomedical Informatics used in this area of research. Link:

Speaker: Miquel Angel Mayer, MD, Senior Researcher, Integrative Biomedical Informatics Group - Research Programme on Biomedical Informatics (GRIB, IMIM/UPF)

Room Online

Thursday, 25th March, 2021, 12:00

New gene explosion in three different organisms regardless of the mechanism of origin

PRBB Computational Genomics Seminar. Chair: Mar Albà (Head of Evolutionary Genomics Group). New genes are constantly created in all organisms by two different mechanisms: gene duplication and de novo gene birth. The first mechanism generates an exact copy of a gene that will diverge from the original. On the other hand, de novo genes are originated from previously non-coding regions producing a completely original gene. Both mechanisms contribute to the generation of new genes, however, their relative contribution to short-time adaptation and long-term evolution is not well understood. In our group, we used a standard methodology to measure the rate of formation and persistence of both, de novo and duplicated genes in three different groups formed by yeasts, insects and vertebrates. We observed that independently of how new genes are formed the number of genes is higher at the species' level and then rapidly declines until a steady-state level. We also studied the purifying selection of groups of genes with polymorphism data and we observed weak purifying selection, that increases over time. Zoom webinar: / Password: 788137

Speaker: Jose Carlos Montañés, Evolutionary Genomics Group - Research Programme on Biomedical Informatics (GRIB, IMIM/UPF)

Room Online

Thursday, 18th February, 2021, 12:00

Multivariate analysis of single-cell and spatial transcriptomic data

PRBB Computational Genomics Seminar. Chair: Robert Castelo (Head of Functional Genomics Group). Recent technological advances in molecular biology allow the sequencing of RNA from individual cells (single-cell RNA-seq) and the joint analysis of gene expression and spatial cellular organization (spatial transcriptomics). Typically, the genes whose expressions are differential between cell states or across the different areas of a tissue slice are identified with univariate (gene-wise) models. However, it may be beneficial to explicitly account for gene dependencies in multivariate statistical models. In this talk, I will show two examples of such models: (i) a graphical model for single-cell RNA-seq and (ii) a co-clustering model for spatial data. I will show how to use the first model to explore the dynamics of transcription factors in development and the second model to jointly identify spatially dependent genes and cell states/types. Finally, I will discuss the advantages of community-based open software development in Bioconductor. Zoom webinar: / Password: 788137

Speaker: Davide Risso, Dept. of Statistical Sciences, Università di Padova, Italy

Room Online

Thursday, 14th January, 2021, 11:00

Neural network potentials for protein folding

Zoom webinar:

Meeting ID: 931 1763 6692 / Password: 160118

Speaker: Maciej Majewski, Computational Science Group - Research Programme on Biomedical Informatics (GRIB, IMIM/UPF)

Room Online

Thursday, 5th November, 2020, 12:00

Systems medicine temporal analysis of patient disease trajectories

Incorporating the temporal dimension into disease comorbidity studies has shown to be crucial for achieving a better understanding of the disease progression. Furthermore, due to the multifactorial nature of human disease, exploring disease associations from different perspectives can provide a global view of the underlying mechanisms and thus, shed more light on their aetiology. In this work, a temporal systems-medicine approach is proposed for identifying time-dependent multimorbidity patterns from patient disease trajectories, by integrating data from electronic health records with genetic and phenotypic information. Specifically, the disease trajectories are clustered using an unsupervised algorithm based on dynamic time warping and three disease similarity metrics: clinical, genetic and phenotypic. The proposed integrative methodology can be applied to any longitudinal cohort and disease of interest. Prostate cancer is, herein, selected as a use case of medical interest to demonstrate, for the first time, the identification of temporal disease associations in different disease spaces. Zoom webinar:

Speaker: Alexia Giannoula, Integrative Biomedical Informatics Group - Research Programme on Biomedical Informatics (GRIB, IMIM/UPF)

Room Online

Friday, 24th April, 2020, 10.30

Soft-matter materials modeling in the data-driven era

Advanced statistical methods are rapidly impregnating many scientific fields, offering new perspectives on long-standing problems. In materials science, data-driven methods are already bearing fruit in various disciplines, such as hard condensed matter or inorganic chemistry, while comparatively little has happened in soft matter. I will describe how we use data-driven methods to leverage molecular simulations in soft matter. We aim at establishing structure-property relationships for complex thermodynamic processes across the chemical space of small molecules. Akin to screening experiments, we devise a high-throughput coarse-grained simulation framework. Coarse-graining is an appealing screening strategy for two main reasons: it significantly reduces the size of chemical space and it can suggest a low-dimensional representation of the structure-property relationship.

I will illustrate these aspects through the passive translocation of small molecules across a phospholipid bilayer, identifying interpretable structure-property relations, as well as recent results on the screening of small molecules to drive phase transitions in lipid mixtures. Further applications of recent machine learning architectures to molecular simulations will be described: representation learning of a free-energy landscape using Gaussian-mixture variational autoencoders, and the systematic backmapping of coarse-grained configurations to an atomistic level using conditional generative adversarial networks.

Speaker: Tristan Bereau, Van't Hoff Institute for Molecular Sciences and Informatics Institute, University of Amsterdam, Netherlands

Room Charles Darwin, PRBB Innner square

Thursday, 16th January, 2020, 12:00

Unravelling the mystery of orphan genes to understand the origins of genetic novelty

What explains the presence of a gene only in the genome of one species and not in any other? Species-specific protein-coding genes, also known as orphans, can arise "from scratch" from previously non-genic loci, through a process known as de novo gene emergence. How exactly the evolutionary transition from non-gene to functional gene unfolds is unclear. Can such de novo emerging genes increase an organism's fitness, and if so how? Orphan genes can also result from extensive sequence divergence of ancestral genes, which can eventually erase all similarity of a gene to its homologues in other species, a process even less well understood than de novo emergence.  I will present novel findings which advance our understanding of both these evolutionary mechanisms and bring us a small step closer to a complete picture of the origins of genetic novelty.

Speaker: Nikos Vakirlis, Benaki Phytopathological Institute, Athens, Greece

Room Aula room 473.10 (4th floor)

Site Information