Please enable JavaScript.
Coggle requires JavaScript to display documents.
Talks (Martin Hemberg: Computation analysis tools for single cell RNAseq…
Talks
-
-
-
-
-
-
-
-
Global Biodata Coalition: Rolf
Paper: Sustaining the big-data ecosystem - Philip E Bourne
F1000 Research Article: Identifying ELIXIR Core Resources
EBI has not encompassed clinical data since the primary focus is research.The closest resources are dbGAP in US and EGA in Europe.
Global Data Infrastructure for the scientific enterprise: Niklas Blomberg
Open data requires infrastructure
It takes much less to maintain knowledge than to generate it.
Research infrastructure: user-focused service (core resource)
Indicators of core resources: Scientific focus, impact, community served, quality of service, legal & funding
Paper: Public data resources as a business model for SMEs, Elixir (Mind The byte, Repositive, The hyve, Ontoforce, General Bioinformatics, eagle genomics,ribocon)
- Gene Curation Coalition: Rebecca Foulger
- Curating genes with respect to clinical research
- Gene-disease association: Genomics England PanelApp
- Variants in patient->Rare variants->Protein coding variants->different variants in parents->how many listed in PanelApp
- PanelApp categorizes gene association with a disease in a tier system
- PanelApp is free
- PanelApp contains information on genes and not the variants.
ClinGen: Marina
- gene-disease validity: accesses the validity of a disease with a gene
- Clinical Genetic Registry
- Volunteer curation
Orphanet: Antonie Marmigon
- Clinical database not a genetic database
- Over 40k genetic test available
- Interoperability between rare disease, genetic test and genes.
- Provide panel of genes for genetic testing.
- Validation process: single paper, internal gene meeting, expert advices, advisory board o genetics.
- ARL3 mutations - Joubert Syndrome paper
Biocuration in industry:Jane Lomex (Scibite), Eagle Genomics, Gentech
- Nebion AG: Biocuration of transcriptomics like RNA seq and microarray from literature manually (Genevistigator-licensed)
- Roche: There is a need for data integration in pharma. Biocuration of clinical biomarkers/technological standards (Pharmacovigilance/Clinical Trials)
- HealX: Databases->Biocuration->ML/Text mining->HealX Knowledgebase
- Eagle Genomics: provides efficient software data handling and analyzing tools for biologists & healthcare.
- Pistoia Alliance: a non-profit initiative to collaborate several pharma companies to speed up the drug-discovery process. One of the project is ontologies mapping (Paxo).
Q&A:
- Manual work: Excel data inconsistency, GO ontology choice, ranking and visualizing results, meta data on results.
- Biocurators should market themselves as something else - Knowledge Engineer/Data Readyers.
Unheard resources:
- APID: DB for PPI experimental evidence
- ClinVar: Variants in clinical data
- GeneWeaver: Curating gene sets
- OBI: can be used as a curation tool
-
FAIR standards: Susanne
- isaTools
- FAIRSharing
Springer Nature: Varsha Khodiyar
- Springer tries to promote free open publication
- Researchers share data in many forms mainly including emails, supplementary materials, excels, database archives.
- Primary challenge in data sharing in Biological science is the organization and visualization of data.
- Top questions of researchers willing to share are 'How to share the data?' and 'what is the policy for this?'.
- http://tinyurl/y8atpy8a
- http://tinyurl/yy4qnqs5
Recommended Interoperability Resource at ELIXIR: Sira
- RIR: A platform that facilitates the analysis of curated core resources.
- Use case in marine meta genomics, rare diseases, human data & plant sciences, Galaxy, Proteomics.
- Integrate resources seemlessly, standardize APIs
- RIRs is a process that integrates different resources.
Genomic Standards Consortium (MIxS): Lynn
- A standard for all sequence data.
- Minimum Information of sequence Information (MIxS)
- BioSamples
Preprints in life science: Naomi Penfold: ASAPbio (scientist-driven non-profit) - bit.ly/EBI-preprints-2019
- Helps speed the publication process.
- Researchers post their work as soon as they have something. The community can then comment and review the content.
- This work can then be published in journals.
- The community should comment more on preprints.
- Other resources: arXiv (physics), bioRXiv (biology), SHARE (humanities).
- Only 2% of biomedical publications of the net are preprints.
- ASAPbio is not just about preprints - they also offer other ways to make biology research faster and transparent.
- asapbio.org/preprint-info
- asapbio-ambassadors
- reimaginereview - transparent review system
- Preprinting in bioinformatics - bioRXiv & arXiv
- EMBL has a preprint channel.
- Similar to open access journals.
- asapbio.org/transpose-preprints
Fritz Roth: Mapping environment-dependent protein networks and functional human
sequence variation
- Systematic mapping of interacting proteins
a. Luck et al, 2019, BioRXiv.
b. Human Reference Interactome (HuRI)
c. Standardize a procedure to monitor organism level protein interactions. Then do the same for different environmental conditions. Changes can be seen but the precise reason for this change is not certain.
- Systematic testing the function of human misense variants
a. Only 2 % of genome is sequenced for misense variants of which most are useless (ClinVar).
b. Fowler & Fields, Nature Methods 2014 - have a map of different variant to see clinical effects: Variant Effect Predictor.
c.Sun et al, BioRXiv, 2019 - pathogenic variant prediction.
d. Weile et al, Mol Sys Bio 2017.
e. MaveDB: Database for sharing variant effect genes.
e. Goal is to create a variant effect map for all human disease genes - about 3970 genes.
f. MutFunc: Good to find loss of function mutations/variations.
g. Mutfunc is a database of mutations occurring in functionally important regions or that are predicted to disrupt protein structure stability, protein interaction interfaces, PTMs, protein translation, conserved regions and regulatory regions.
-
CORBEL: Clinical Genomics
Jan Willem Boiten & Menno de Vries
- tranSMART & cBioPortal
- problems in IT in translational reasearch - multimodal collaboration
- What is needed is a standardized ecosystem
- Created an app store for translational research using available free tools.
- TranSMART - a clinical genomics integration platform
- TranSMART - originally developed by Johnson & Johnson. Then was made open source because it was hard to maintain.
- cBioPortal - a gene-centric view of clinical data.
- cBioPortal - visualization, analysis and download of large scale cancer genomics data; open source
- i2b2 - open source clinical research warehouse
- i2b2 and tranSMART are merged now
- XNAT - open source to share image data and analysis.
- HEALTH-RI is a collaboration effort with major players in the field.
- tranSMART can be downloaded in local machine for use.
- Data in the same clinical trial is compared. In order to compare different clinical trials, they need to use the same ontologies.
- The patient data in tranSMART can't be linked to the genetic data in cBioPortal. They've to be treated separately.
ML in Image analysis - Anna Krushuk
- Boundary segmentation problem - it is essential to separate different organelles in electro-
microscopy images.
- It is efficient to predict affinities than to predict boundaries
- Cell segmentation problem
- Scaling up is an issue
- Identify boundaries for pixels-> Merge to form Superpixels -> merge further to agglomeration
- Big data set - Neurons images
- Segmentation toolkit- ilastik
- Add prior knowledge to NN models to train the model to identify some features that are missing in training data.
- Multi-modal data: classify cells in whole organisms - gene classification in different cells. (Platy browser)
- NN model: U-net for segmentation (supervised)
- NN model: Y-net for segmentation (unsupervised)
- ilastik - open source
- Model zoo
MOL2019 Workshop
Lourdes:
- PK - what happens to the drug
- PD - what the drug does to the system
- PBPK - physiological dynamics of the drug
- QSP - fills the gap between drug-target models and drug-discovery
Alessandra Gaeta: Psychiatry consortium
Challenges:
- Lack of new validates drug targets
- High failure rates in clinical trials
Aled Edwards (Univ of Toronto)
- We don't understand the biology of many diseases let alone treat them.
- So, pharma's make a lot of money when they discover a drug.
- Science resist change and is redundant.
- Opportunities lies in the kinase that have no established function.
- Wellcome Trust and SGC are exploring this domain.
- Drug testing fail results are not reported by pharma companies and drugs are tried as last option for patients repeatedly.
- SGC aims to develop this ecosystem to understand protein functions.
- Agora open science trust: non-profit pharma company.
- M4K Pharma Inc.: Virtual biotech model/no patent.
- Regulatory exclusivity vs Patents
- Ideation and implementation of an open science drug discovery business model - M4K Pharma.
- M4K/M4KND/M4ID/M4??
Irene: Gene expression
- Microarray and bulk RNA-Seq have been around since 2000.
- Single cell expression has started recently around 2014.
- Significance: differential expression in response to treatment, location and time.
- ExpressionAtlas: Data->Archive->Analyse data->Present data
- Challenges: data integration from different sources or experiment types.
- Solutions: use ontologies, standard tools, meta analysis and batch correction.
- Comparison between bulk and single cell expression data will inform about specific cell type dynamics.
- Application: Identify genes in diseases (identify biological variance by comparing normal vs disease)
- Application: meta-analysis of ND (batch correction of available studies and proceed from there).
- Future: cell type deconvolution of phenotypes and diseases (identify high expression genes in low abundant cells in tissues)
- Future: cell type abundance in different tissues (spatially resolved transcriptomics integrated with single cell data) - in gut cell atlas project.
- Future: how similar are cell types in homologous tissues between different species (other single cell atlases available).
- Questions: In cell type deconvolution how coupled are gene expression patterns and celltypes (for instance in diseases)?
14: Questions: How to validate batch correction? - one approach is to look for the appropriate gene expression patterns.
- Questions: How do you annotate cell types? - annotations are taken from data generators but also other tools to annotate or manual annotations.
- Questions: How do you resolve epigenetic components and chromatic-accessible components? - not known.
Elixir open day:
- Elixir is a research infrastructure aimed at safe guarding research data.
- Provides networking opportunities - industry and SME.
- Hosts job vacancies.
- Training portal - TeSS.
Chuck Cook (Global Biodata Coalation)
- Ensure sustainability of biodata infrastructure
Elixir Tools
BioConda
Galaxy tools
- Training: TIaaS
Elixir Compute
Oliver Stegle: Genomic statistics
- Gene-environment relation study.
- Resources: biobank
- gene + Env + GxE = phenotype
- plot phenotype(BMI) vs risk alleles: different lifestyle changes the slope of the line
- Research involves combining different types of environmental conditions and identifying genetic associations.
- biobank has several GxE data.
- Questions: which environmental variables drive GxE?
- Questions: genome-wide GxE analysis of traits?
- Questions: GxE to molecular traits?
- Resources: GTeX has expression data on different cell/tissue types.
- Questions: how does the genetic regulation change based on GxE associations?
- Research: eQTL discovery across differentiation stages.
- Research: quantifying genetics (allele-specific) at the single-cell level.
- Questions: what is the genetic profile at different stages while also accounting for individual changes.
- Challenge: how are GxE and ExG related? (knowing causation is tricky)
- Challenge: How to get reliable answers on fuzzy factors (how much TV do you watch a day)?
-
Biocuration for the 100k genome projects & beyond: Ellen McDonagh (Genomics England)
- Genomics England took over the 100K genomics project: whole genome seq. of rare diseases or cancer->identify variants->set the platform for NGS
- GeCIP: research community usage of the 100K data
- Biocuration at Genomics England
- PanelApp: evidences are available here.
- PanelApp integrates several resources like OpenTargets, Disgenet, VeP, Ensembl, etc.
- Pharmacogenomics: how does genetics affect drug response - curate variants relevant for different diseases.
-
