Aided by a single slide all delegates receive 2-3 minutes of speaking time to introduce themselves and their research. This will promote future networking opportunities and indepth scientific discussion.

Submission guidelines

Submission website: https://www.easychair.org/conferences/?conf=jpiismb2014. Submissions will be accepted until 5pm Saturday July 12.

Please include the following information in the abstract box on the submission website:

• a short bio about you and your research to share with participants (maximum 200 words)

• any topic recommendations for the round table discussions

• whether you will attend the JPI lunch following the workshop to continue round table discussions
  

To facilitate discussions we will organize a moderated round table discussion in which each table will discuss a certain subject. The goal of the discussion is to jointly identify a set of lessons or insights that will then be shared with the rest of the delegates during the plenary feedback session.
We will define a selected list of topics and assign delegates to the subjects according to their interests.
Each group will discuss the subject at hand and share their experiences. Each group has a discussion leader and someone taking minutes. Potential subjects are:
● Grant writing
● Hiring and supervising
● Setting up collaborations
● Leading a team
● The role of social networks/media in maximizing research impact
Suggestions for additional topics will be solicited prior to the workshop. The round table discussions will continue into lunch. Each participant can have an individual bill for lunch and mains cost $10-20

Is there a strong case for using commercial software in a bioinformatics core or are free open source alternatives not only viable but preferable? Here two differing bioinformatics core facilities will make a case for their choices in both hardware and software infrastructure. The  discussion will focus on practicality, scientific validity, ideology and reproducibility, but will also cover constraints of budget and time as well as the dynamics between users and core facility staff.

Alastair Kerr, running a core on a budget, maintains his own servers and has made the decision to empower users to do at least a basic amount of analysis themselves. He is an advocate for free open source software and data, finding that many such solutions are preferable to commercial alternatives. In this talk he will detail such software that has been used for interfacing with users as well as some areas where commercial or semi-commercial products are still used. Like most core facilities, he has had to deal with the influx of large quantities of deep-sequencing data and will describe the associated budget-focused choices in hardware that were made.

Michael Poidinger has previous experience in pharma Discovery IT and currently runs a core group for an academic/governmental translational immunology research institute.  The group is very well funded by central funds with little need for cost recovery and focusses more on a collaborative rather than service model. At the same time his team of 8 handles 200+ people in 28 groups together with numerous external partnerships, so throughput and efficiency are also of concern.  The group is founded on two commercial products, Pipeline Pilot from Accelrys and Spotfire from Tibco/Perkin Elmer.  In this talk he will describe how the software first seen in the pharma environment has been brought over to an academic one, and the improvements in productivity that these tools provide.

The discussion will pick up on the themes introduced by the speakers. We anticipate that participants will represent a diverse cross-section of bioinformatics groups with varying levels of resources both in terms of staffing and operating budgets. Some groups will have limited capacity to implement comprehensive bioinformatics infrastructure and are perhaps more likely to deploy open source or freely available software and analysis workflows developed by the wider academic community. These groups may have developed innovative approaches to overcome the constraints they face. Conversely, organizations that have been able to invest in enterprise-level offerings from bioinformatics software vendors will be able to provide an insight into how well these solutions work in practice that should be of interest to those making “buy versus build” decisions. In between these two extremes will be groups that have sufficient personnel to develop in-house solutions but where there may be significant duplication of effort to achieve the same end goals, e.g. many groups have developed their own workflow engines, sequencing LIMS or systems for tracking analysis projects but did the specific requirements of their institute or organization really justify this? In the open forum discussion we will explore the choices made by different groups.

This part of the workshop will be a structured discussion which will cover several different topics spanning the technical, scientific and managerial aspects of core facilities.  In contrast to the previous session there will be no named invited speakers and the entire session will be based around moderated group discussions.

The rationale for this session was constructed around observations made at previous workshops, namely that:

1. The sessions for a workshop are planned several months in advance of the conference, and it is often the case that new technologies arise in the intervening time which would merit discussion but are then difficult to include.

2. Managers of core facilities often have short topics on which they would like to solicit the opinion of the bioinfo-core community, but the previous structure of the workshops has not allowed for this.

3. The discussions in workshops are often led by a more vocal subgroup of the whole community and we wanted to find ways to allow more people to become involved.

The aim of the session therefore is to end up with a set of community submitted late breaking topics for discussion which would be presented by the moderators in the workshop.  Each topic would be the subject of short, focussed discussion (5 - 10 minutes) so that we can try to cover a range of different areas of core facility management in order to address the interests of as many people as possible.  Within the constrained time we may not manage a complete coverage of an individual topic,  but those which promote most discussion could be taken forward for online coverage or be included in future conference calls or workshops as more major items.

The practical way in which we envisage this session to work would be that several weeks before the meeting we would put out a request to the bioinfo-core mailing list asking for submissions of short topics which they would find interesting. The topics could be a simple question, a suggestion of an area for discussion or a specific observation which they thought was interesting. Topics can be supplemented with a single explanatory powerpoint slide if needed. All submissions will be presented anonymously unless people specifically say that they are happy to have their name attached.

Around a fortnight before the meeting the workshop organisers will get together to go through the submissions. They will put them into rough groupings based on similar areas of interest and will deduplicate and rationalise them to try to tie together similar themes so that discussion can more naturally flow from one topic to another. They then finally put the deduplicated topics in order based on their popularity.  The finalised topics for the session could then be advertised at this point.

At the meeting two moderators will be responsible for managing the session and will work their way through the questions putting them out to group discussion and deciding when enough time has been spent on a topic to move on to the next one. We would aim for each topic to get around 5 minutes of discussion

After the session the whole set of discussion topics and associated notes from the discussion will be put up onto our bioinfo-core wiki so that and topics which were not sufficiently well covered in the meeting could be continued online.

Dr. Love will speak on how to produce content for an open online course, the mechanics of running a course, how such courses can (or cannot) be used to teach bioinformatics, and what resources are available for someone interested in creating an online course.

Dr. Good will speak on online games for bioinformatics and how they can (or cannot) be used to teach bioinformatics.

This talk will review the state of the MOOC universe as experienced by those seeking to study computational biology and related fields. It will include assessments of the current coverage of key topics, the quality and consistency of course offerings, and specific opportunities for improvement in both. Particular attention will be paid to the strengths and weaknesses of the online format with regard to achieving proficiency at various levels in bioinformatics and computational biology.

The panel will include the 3 speakers plus a remote speaker (to test online learning systems!), and will debate the merits and pitfalls of online learning for bioinformatics. Do online learning mechanisms adequately teach bioinformatics, or must online learning be supplemented with other modes of learning? Where does online learning in bioinformatics need to go?

Bioconductor is a collection of more than 700 individually code-reviewed software package, hundreds more annotation and experiment data packages, and specialized data structures for various domains.   This talk will provide a high-level overview of Bioconductor in the domains of gene expression, DNA variant calling, flow cytometry, proteomics and metabolomics.  This talk will provide users with an up-to-date overview of Bioconductor’s offerings and recent developments in these diverse domains.  It will also review options available to users wishing to analyze public data from souces such as The Cancer Genome Atlas, the Gene Expression Omnibus, and ArrayExpress, and data distributed by Bioconductor itself.

Bioconductor provides a wealth of genomic annotation data as pre-compiled packages: gene-centric, transcript-centric, functional, and technology-specific. However as the size and richness of genomic annotations increases, such as generated by the ENCODE project, a more flexible and scalable approach is needed.  AnnotationHub is a database-driven, client-server architecture that meets these needs and is more convenient for creators and users of genomic annotation.  Annotations are added through simple “recipes” linked to versioned data and metadata; users can retrieve snap shots from any point in the history of a particular data resource, or search for data resources based on any of the associated metadata.  A beta version already provides more than 10,000 ENCODE tracks as prepared Bioconductor objects, and we expect AnnotationHub to overtake the current package-based model of providing annotation and experimental data in Bioconductor over the next several years.  This presentation will introduce the architecture and usage of AnnotationHub, for both users and contributors.

We discuss how to decompose data and algorithm elements so that available hardware configurations, be they modest or vast, can be used to maximize throughput of bioinformatic analyses.  We prefer algorithm and data structure designs that do not require developers to make strict advance commitments regarding memory requirements and degree of parallelism achievable.  Instead, we want something close to run-time determination of suitable approaches to resource utilization.  Ultimately such determinations should be autonomous; at present, user-driven selection of subproblem size and compute resource size and structure are needed, but the software and the software interface should allow great flexibility in mapping the solution process to the resources according to these selections.  We will discuss a number of use cases, including the evaluation of different procedures for inference on eQTL.  The required computations are intensive and involve substantial volumes of experimental and annotation data. An Amazon EC2 machine instance (AMI) has been created with all necessary software and data to illustrate the entire procedure, and we describe how to use the open StarCluster configuration tools to create and use clusters based on this AMI to solve various genomic discovery and interpretation problems with Bioconductor. 

Bioconductor contains a number of software and annotation packages which simplify an RNA-Seq analysis workflow. This session will cover a standard pipeline: obtaining gene models and counting RNA-Seq reads in genes and exons; performing quality assessment and exploratory data analysis using transformed counts; performing differential expression at the gene or exon level; and finally annotating results and facilitation of downstream analysis.  Additionally, new DESeq2 functionality will be demonstrated which allows for moderated effect size estimates for an individual treatment relative to a number of other treatments.

Academics often talk of the inefficiencies in current peer-review systems, and there have been suggestions for reform in various ways in past years. Recently a number of journals and initiatives have looked at new approaches to peer review to address these concerns - journal policies to reduce the number of iterations of review aiming for speed of publication, open peer review to increase transparency of the process, post publication peer review with new ways to surface quality research, or de-coupled review services that move responsibility for peer review from the publishing journal to the research community.

This presentation will explore some of these new approaches,as well as those with established variations on the traditional model such as Biology Direct (a journal edited by Eugene Koonin, David Lipman and Laura Landweber that since launch in 2006 has aimed to return the responsibility in peer review to the authors and reviewers, by allowing authors to choose reviewers, and by including named reviewer reports in the published manuscript as a guide to the paper as seen). Transitioning into our next section on reproducibility, we ask whether peer-review of software and data is needed and explore whether cloud computing might provide the infrastructure needed for a more painless peer-review of software and data.

Good data stewardship starts with good plans. This includes good description of experimental plans. In spite of the big data buzz, for most scientists, sound experimental setup is at the core of their work, with good level of replication giving strength to evidence. Yet very few make the most of experimental design information for managing, reporting but also reviewing and publishing scientific data. In this session, we would like to explore how simple principles derived from experimental design practice can greatly enhance data generation, data quality, data reporting, data review and data publication. The experience garnered over an experiment in reproducibility will also touch on issues associated with reporting findings.

In the big data era, our funding bodies are starting to require a proper data stewardship plan for any project that produces data by public funding. What does this mean for the bioinformatics community? What does this mean for getting credit for the valuable work that we do to produce, preprocess and interpret data? What tools will we have available to increase the value of data, especially when data and method sharing will become a must. In this session Barend Mons addresses issues involved with the new trends in data publishing, data models and tools that support the bioinformatician, and the opportunities these present for biological research.

This final session will bring together a panel of representatives from research and community initiatives to present a range of new approaches to peer review and publishing of software including Kaitlin Thaney (as a panel host), Carole Goble, and Phil Bourne. This panel will explore questions around software publishing and reuse, giving the audience the opportunity to participate. Questions to be explored include but are not limited to:

• The practice of assigning identifiers to code so that it can be integrated into scholarly publications is becoming more of a reality, but what level of responsibility do reviewers now have for ensuring that code is sound?
• What information is needed for the code to be picked up, forked and run by someone else outside of their lab?
• What are the minimal metadata fields needed? The infrastructure (documentation, APIs, etc)?
• How can we best surface information immediately of use to the researcher that gives them the necessary information to understand, use and build on the code made available?
• In terms of code and best practice for reuse, what can we start doing right now? What do we need to aim for in the future?