Seminars, events & talks

Friday, 23th November, 2018, 10.00 - 11.00

Deep Reinforcement Learning for Partially Observable Environments

Many real-world sequential decision-making problems are partially observable by nature and the environmental model is often unknown. Examples include visual occlusions, unobserved latent causes like in healthcare or when we rely on noisy sensors. Consequently, there is a great need for reinforcement learning methods that can tackle such problems given only a stream of observations.  In this talk, I will briefly present the two fundamental approaches how partial observability can be tackled when we want to learn a policy, namely memory or inference. Subsequently, I will present our recently proposed algorithm "Deep Variational Reinforcement  Learning" (DVRL) which combines learning a model with particle filtering to allow the agent to reason more effectively about the unobserved state of the environment.

Speaker: Maximilian Igl, Department of Computer Science, University of Oxford, UK

Room Marie Curie, PRBB Innner square

Thursday, 4th October, 2018, 12:00-13:00

Is ribosome profiling better than RNA-seq for estimating protein abundances? A case study in Saccharomyces cerevisiae in normal and oxidative stress conditions

Yeast cells respond dynamically to different environments; often this response involves adjusting the abundance of different proteins. In many published experiments, this cellular response is studied by proxy-measuring variations in transcript abundance across different conditions using high throughput RNA sequencing. However, variations in transcript abundance do not always reflect changes in protein abundance. Ribosome profiling, or Ribo-seq, specifically targets only the ribosome-protected fragments of RNA; as there is pervasive translation of ribosome-assocaiated RNAs, this would be expected to provide a more accurate view of changes at the protein level than RNA-seq. We performed a differential gene expression analysis of oxidative stress in baker's yeast using RNA-seq, Ribo-seq, and liquid chromatography mass spectrometry. A subset of genes involved in oxidation-reduction processes is detected by RNA-seq and Ribo-seq, but RNA-seq also identifies many genes which are not significant in the Ribo-seq analysis which suggests that significant changes in mRNA abundance do not always result in different protein abundance. Furthermore, the proteomics data more closely resemble the Ribo-seq results than RNA-seq. Our findings demonstrate that there could be limitations to using RNA-seq when making inferences about changes in protein abundance and the potential of Ribo-seq to fill this gap.

Speaker: Will Blevins, Evolutionary Genomics, GRIB (IMIM)

Room Aula 473.10 (PRBB, 4th floor)

Thursday, 26th April, 2018, 12.00:13.00

Understanding Complex Systems Using High-Dimensional Neural Network Potentials

Speaker: Jörg Behler, Theoretical Chemistry Institute of Physical Chemistry, Georg-August-University Göttingen, Germany

Room Xipre 173.06 (PRBB, 1st floor)

Sunday, 15th April, 2018

3rd European Conference on Translational Bioinformatics: Biomedical Big Data Supporting Precision Medicine

The conference will be held at Barcelona Biomedical Research Park (PRBB) on April 16-17th 2018. This meeting follows on from the success of the European Conference on Translational Bioinformatics (ECTB) series hold in 2016 at the University of Copenhagen, Denmark; and in 2017 at the European Bioinformatics Institute, UK. As in past editions, the 2018 meeting will bring together scientists, developers, and entrepreneurs who are interested in translating genomics and bioinformatics research into healthcare tools and services. The meeting will be relevant for researchers, computational biologists and entrepreneurs from start-ups and established companies interested in epidemiology, cancer genomics and precision medicine. Registration is open here.  For further information of the event, speakers and agenda, please visit the website

Room PRBB Auditorium

Thursday, 12th April, 2018, 11:00 - 12:00

Artificial intelligence: From predictions to sequential decision making

Speaker: Anders Jonsson, AI&ML Research group, Information and Communication Technologies Dep, UPF

Room Xipre 173.06 (PRBB, 1st floor)

Tuesday, 20th March, 2018, 11:00 - 12.00

Thesis defense: "Next generation of informatics tools for big data analytics in drug discovery".

Speaker: Maria del Carmen Carrascosa Baena, Systems Pharmacology, GRIB (IMIM)

Room 61.127, UPF Campus del Mar (Dr. Aiguader 80, 1st Floor)

Wednesday, 7th February, 2018, 12:00-13.00

Measuring ribosome profiling at isoform level: a step towards unveiling alternative splicing funcional impact

The alternative processing of transcribed genomic loci through alternative transcript initiation, splicing and polyadenylation, is an intermediate step in mRNA maturation, between transcription and translation. These mechanisms produce different mature mRNA transcripts and determine the transcript repertoire of cells. There is evidence showing that the differential production of transcript isoforms, especially through the mechanism of alternative splicing, is crucial in multiple biological processes such as cell differentiation, acquisition of tissue-specific functions, DNA repair, as well as in multiple pathologies, including cancer. This has been exhaustively shown at RNA level but it remains elusive at protein level. Sequencing of ribosome-protected RNA fragments, or ribosome profiling, provides detailed information on the transcripts being translated in the cell. In this work, we have developed a pipeline for the quantification of individual transcript coding sequences (CDSs) from ribosome profiling using both RNA-seq and Ribo-seq data from multiple datasets. Moreover, we investigated at wich extend the isoforms that can be detected with ribosome profiling show further evidence of translation.

Speaker: Marina Reixachs, Computational RNA Biology, GRIB (IMIM/UPF)

Room Aula 473.10 (PRBB, 4th floor)

Wednesday, 10th January, 2018, 12:00 - 13:00

The impact of ribosome profiling in the observed conservation patterns of lncRNAs

Several recent studies have noted that a large fraction of long non-coding RNAs (lncRNAs) associate with ribosomes. Deep sequencing of ribosome-protected fragments, or ribosome profiling, provides detailed information on the regions that are translated in a transcript and revealed that many lncRNAs contain translated open reading frames with 3-nucleotide periodicity. However, ribosomes are not specifically selected during the biochemical isolation procedure, and non-ribosomal ribonucleoprotein complexes are also present in these samples. These regions show no periodicity and are highly localized, so ribosome profiling can also identify different RNA-protein complexes with high resolution. In our study we examine the patterns of ribosome profiling in mouse hippocampus to find enrichment of translated open reading frames and ribonucleoprotein complexes in different categories of lncRNAs. We identify conserved regions at sequence level between mouse and human to find putative functional regions in lncRNAs. We examine how the overlap with protein-coding transcripts and promoter signals also affects the conservation of lncRNA sequences.

Speaker: Jorge Ruiz, Evolutionary Genomics, GRIB (IMIM)

Room Aula 473.10 (4th floor)

Wednesday, 13th December, 2017, 12:00 - 13.00

Genetic linkage analysis of heritable pulmonary arterial hypertension in a large pedigree identifies candidate modulators of reduced penetrance

Large-scale genetic profiling and clinical sequencing are revealing an increasing number of carriers of disease-causing mutations who do not develop the disease phenotype. This characteristic is clinically reported as a genetic disorder of reduced or incomplete penetrance. Several mechanisms have been proposed to explain reduced penetrance, such as the molecular context of mutations, patient characteristics, such as age or sex, as well as specific environmental conditions that delay or trigger the disease onset. The phenomenon of reduced penetrance constitutes a major challenge in the field of genetic diagnosis and counselling because phenotypes no longer unambiguously exhibit underlying genotypes. Nevertheless, its existence also provides new opportunities to learn how genotypes shape phenotypes. In this talk, I will describe our efforts using linkage analysis to find a genetic modifier that explains the reduced penetrance in a particular genetic disorder: heritable pulmonary arterial hypertension. The results from linkage will be further discussed regarding evidence on haplotype prediction, functional enrichment tests as well other functional genomics tools. These steps are required to narrow down the list of potential candidates mapping the modifier and eventually to hypothesize about a particular genetic mechanism underlying reduced penetrance.

Speaker: Pau Puigdevall, Functional Genomics, GRIB (UPF)

Room Aula 473.10 (PRBB, 4th floor)

Wednesday, 8th November, 2017, 12:00 - 13.00

PsyGeNET: a knowledge resource on psychiatric diseases and their genes

Psychiatric disorders constitute one of the main causes of disability worldwide. During the past years, considerable research on the genetic architecture of such diseases has been conducted, although little understanding of their etiology has been achieved. The difficulty to access up-to-date, relevant genotype-phenotype information has hampered the application of this wealth of knowledge to translational research and clinical practice in order to improve diagnosis and treatment of psychiatric patients. PsyGeNET has been developed with the aim of supporting research on the genetic architecture of psychiatric diseases, by providing integrated and structured accessibility to their genotype-phenotype association data, together with a set of analysis and visualization tools. PsyGeNET contains up-to-date information on genes associated with mood disorders (depression, bipolar disorder), psychosis (schizophrenia) and substance use disorders (alcohol, cannabis and cocaine use disorders, substance-induced depressive disorder and psychoses). The current version of PsyGeNET (version 2.0) contains 3,771 associations between 1,549 genes and 117 psychiatric disease concepts. PsyGeNET offers several metrics for the prioritization of the information, including the Evidence Index that takes into account the negative evidence found in publications for a particular gene-disease association. The PsyGeNET database has been developed by extracting gene-disease associations from the literature with the text mining tool BeFree (, followed by a process of curation by a team of 22 domain experts. A web-based annotation tool supported the curation process. The data is accessible through a web interface ( and the psygenet2r Bioconductor package ( Moreover, the psygenet2r package implements several functions to visualize and analyze the PsyGeNET data in a clear and meaningful way and allows performing comorbidity analysis based on shared genes. Due to its special focus on psychiatric diseases and comprehensiveness, PsyGeNET represents a valuable resource for the analysis of the molecular underpinning of psychiatric disorders and their comorbidities.

Speaker: Alba Gutiérrez-Sacristán, Integrative Biomedical Informatics, GRIB (IMIM/UPF)

Room Aula 473.10 (PRBB, 4th floor)

Site Information