275 research outputs found
datacarpentry/wrangling-genomics: Data Carpentry: Genomics data wrangling and processing, June 2019
No full text<p>Data Carpentry lesson to learn how to use command-line tools to perform quality control, align reads to a reference genome, and identify and visualize between-sample variation.</p>
Managing the Data Ecosystem in Life Sciences through Bioinformatics
No full textInformation is key in producing knowledge. We undoubtedly live in time of massive information; particularly in Science, there is an abundance of research published and otherwise disseminated daily. However, volume is not always a positive trend as useful information can be easily lost within the multitudes of similar, but not necessarily relevant, information. Moreover, much of the research produced is hidden behind subscription walls, which severely limit access for the wider audience. This is particularly true in Life Sciences which have become massive in scale through the widespread use of NGS sequencing machines among others. The simultaneous growth of data collection techniques along with data aggregation and mining algorithms provides an unprecedented opportunity for rapid knowledge discovery. Integrated platforms combined with bioinformatics solutions can pave the way towards horizontal studies and novel opportunities. Specifically, the construction of domain-specific platforms can lead to integrated solutions and multiple views of the same issue, by acting as a nexus of knowledge for the particular domain / problem it addresses, and by providing novel opportunities for systematic integration of different data, thus facilitating horizontal studies. In essence, they can a form a framework for Bioinformatics analysis that (a) enables the re-analysis of old data with new methods, (b) facilitates data remixing and data integration for increased resolution and (c) allows for algorithmic and data-driven hypothesis generation.Poster presented at 8th RDA Plenary, Denver, Colorado, USA, 15-17.09.201
Delivering hands-on training for Machine Learning on omics data through GALAXY
No full textA short presentation done in the context of the GOBLET AGM 2020, offering an overview of the online training for Machine Learning that took place using the Galaxy infrastructure
FAIR for Machine Learning; Building on the Lessons from FAIR Software
No full text<p>Ensuring that data are FAIR is nowadays a clear expectation across all science domains, as a result of many years of global efforts. Research software, has only just started to receive the same level of attention in recent years, with targeted actions towards the definition of the FAIR principles as applied to research software, as well as concerted efforts around reproducibility, quality, and sustainability. Given the rapid rise of ML as a key technology across all science domains, it is important to build on our collective experience, and already start addressing the challenges ahead of us, towards making ML FAIR.</p>
<p>These are slides that were presented at the <a href="https://zbmed.github.io/damalos/" target="_blank" rel="noopener">4th Workshop on Metadata and Research (objects) Management for Linked Open Science - DaMaLOS 2024</a>, co-located with <a href="https://2024.eswc-conferences.org/" target="_blank" rel="noopener">ESWC</a> on 26th May 2024, Hersonissos, Crete, Greece.</p>
Machine learning in Biology: Establishing Standards
No full textThe presentation of the DOME Recommendations, that was delivered in the context of the "Machine Learning good practices" workshop (NTB-W01 – ECCB2022 at ECCB 2022, on behalf of the ELIXIR Machine Learning Focus Group
The unappreciated role of k-mers in bioinformatics
No full textThis is a talk given in the context of the BSC Life Sessions
Abstract
k-mers are used on a daily basis in bioinformatics. Although they have existed at the core of several popular tools for genome assembly for quite some time, until recently they have been woefully underutilized. Although k-mer counting is simple and straightforward, it becomes a real challenge when attempting to deal with the huge amounts of data generated in high-throughput sequencing. However, having a simple representation of the actual data with few degrees of freedom (i.e. the k-value and the 4 letters – when dealing with nucleotide sequences), does provide the perfect opportunity to investigate novel mixes of methods and techniques derived from various fields. In that context, the real challenge is to map the biological questions to a corresponding modelling approach. Such examples could be the application of Gödel numbering as a means of transforming the search space for sequence similarity, application of pruned trees and entropy for identifying novel features in sequences, and binning methods for metagenomics classification
Skills and Training in EOSC
No full textPresentation at the Open Science Days in Greece event, organized by ATHENA RC on October 21st, 2021. Highlights the EOSC Skills and Training Working Group output of 2020, as well as the next steps organized under the three Task Forces of the new EOSC Advisory Board
FAIR Research Software as the catalyst for trustworthy AI in Life Sciences
No full textAs a result of many years of global efforts, ensuring that data are Findable, Accessible, Interoperable
and Reusable (also known as FAIR) is nowadays a clear expectation across all science domains. While
data and data management have been the primary focus across many activities, research software has
only recently started getting similar attention. As a result, a coordinated effort by the wider community
allowed to redefine and extend the FAIR principles to research software, with similar activities now in
progress aiming to enhance reproducibility, quality assurance, and long-term sustainability in software
development.
At the same time, we see the emergence of the field of artificial intelligence (AI) and machine learning
(ML) as a key technology impacting all sciences. As AI algorithms and models become increasingly
integrated into scientific workflows, there is an urgent need to maintain high standards for research
software, with the reliability and quality of the underlying software being of primary concern.
High-quality research software is definitely a key catalyst in that direction. In this context, “quality”
involves not only creating robust and efficient algorithms, but also implementing rigorous quality
control processes throughout the software lifecycle. There are several initiatives (such as ReSA, Turing
Way and SSI) that are making available best practices, guidelines and recommendations on research
software, from design and coding to testing and deployment, as well as major funded projects (such
as EVERSE).
Another key aspect is around benchmarking, as it serves as a critical tool for evaluating performance,
scalability, and generalizability of AI solutions across diverse datasets and use cases. In order to
effectively run a benchmarking process, it is essential to establish standardized benchmarks and
evaluations protocols, as well as the respective underlying services and infrastructure to facilitate
this. In both cases, input and direct involvement of the respective community is essential, in order to
fostering transparency and comparability in AI research.
Finally, beyond the technical aspects, there is a clear need for a coherent effort towards the interpretation
of the actual FAIR principles for ML. Some efforts already exist, such as the RDA FAIR4ML interest
group, as well as the efforts under the AI4EOSC project and the ELIXIR infrastructure. However, we still
have some way to go, and direct community involvement is critical to ensure both wide adoption and
ultimately uptake of these practices.Book of abstracts: 5th Belgrade Bioinformatics Conference, Serbia, Belgrade,17-20 june 2024
ELIXIR-GR Training Activities: ELIXIR-GR All Hand Meeting 2020
No full textAn update of the Training activities funded by and/or relevant to ELIXIR-GR. The presentation was delivered during the ELIXIR-GR All Hands Meeting 2020
- …
