Access Keys:
Skip to content (Access Key - 0)

Objective 2.3 GCP-iPlant Meeting 201101

Minutes of the IBP Objective 2.2 workshop held at Marriott Hotel, San Diego, 12-13 January 2011

Present: Scott Chapman, Chunlin He, Hector Sanchez, Fred van Eeuwijk , Naim Matasci, Jiankang Wang, Mark Dieters, Jean-Marcel Ribaut, Graham McLaren, Delphine Fleury, Gouyuo Ye, Trushar Shah, Jeff Ehlers (pm 12th), Alain Charcosset (pm 12th), Laurence Moreau (pm 12th), Rajeev Varshney (13th), Arron Carter (13th), Sonya Lowry (13th by Phone), Jose Crossa (13th by Phone),
Apologies: Arllet Portugal, Abhishek Rathore, Chengzhi Liang, Fred Okono

Summary Conclusions and Action Points

  1. General
  • A lot of progress has been made by the 2.2 team on individual tools which now need to be integrated into a stand-alone suite or a web based workflow system.
  • Some gaps emerged:
    • a lack of decision support tools to follow QTL mapping,
    • no clear vision of how to use simulation tools for tactical questions in the configurable workflow system,
    • final decisions on how to manage genotyping data,
    • no decisions from iPlant on how to deploy databases and which databases to deploy for use in the CWS.
  • Mark agreed to convene the simulation team to consider what simulation tools can be integrated and how.
  • Fred to initiate discussions on how to provide decision support for use of detected QTL.
  • Xavier and Delphine to define a user group to work with Alain on the testing of OptiMAS. First prototype ready at the end of February, once algorithm are verified.
  • Jean-Marcel to consider how to be more proactive with the DOE community and to consider inviting Graeme hammer and Bill Beavis to serve on the SiMAC committee.
  • Naim will consider how to provide an isolated environment for collaborators to develop their own JASON driven pipelines and interfaces. He will consider if it will be possible for the CI to recognize local installations of R and use those to execute the scripts.
  1. Specific Applications
  • Trushar will consider how to integrate several approaches to cross prediction including the classical deterministic approach, and approach based on Alain's tools and approaches based on simulation.
  • Trushar will look at the specifications of MOSEL and see what needs to be added to MBDT to complete the functionality.
  • Trushar to lead the discussion for identifying best genotypic DB. Germinate will be explored, Flapjack format seems an appropriate format for exchange of molecular data but we need efficient storage and database access (Decision to be made before our IBP annual meeting)
  • Fred to work in close collaboration with Naim to have at least the variance analysis available through the iPlant Cyber Infrastructure by June.
  • Guoyou will develop GUI for the different R scripts developed by Fred for: (Variance, GxE, QTL and QTLxE analysis)
  • Several teams have developed R scripts and there is an urgent need to centralize and make the information available to all partners. Delphine will compile a R script inventory.
  1. Project Workplan and Timeline
  • The full team to prepare logframe and timelines by 24th January following the development model presented by Jean-Marcel and the templates to be provided by Fred Okono with the following lead members:
  • Application 1. Workbench Administration and Configuration (Sanchez_)_
  • Application 2. The Breeding manager** The breeding manager (_Sanchez)_
    • The genealogy Manager (_Gupyou Ye)_
    • The Query Tools (_Chengzhi Liang)_
  • Application 3. Field Trial management system** IB Fieldbook (_Sanchez)_
    • Environmental Characterization (_McLaren/Hyman)_
  • Application 4. Genotypic Data Management System (_Trushar Shah)_
  • Application 5: The Analytical Pipeline (_Fred van Eeuwijk)_
  • Application 6: Decision Support Tools** Breeding applications MBDT, MOSEL and MABC (_Trushar)_
    • Marker Assisted Recurrent Selection (_Alain Charcosset)_
    • Cross Prediction (_Trushar Shah)_
    • Simulation Tools (_Mark Dieters)_
  1. IBP Annual Meeting and Training Courses
    We had discussion about the timing of the IBP annual meeting and a training course that Fred was organizing. Unfortunately the dates we discussed clash with some holidays in Europe so we have been working with Fred to select dates which are suitable.
  • The IBP annual meeting will be held from June 1st to 3rd in Wageningen, Netherlands.
  • There will be a course on basic statistics for breeders from 6th to the 10th June
  • There is a tentative plan for a advanced course on selection index theory on 15th and 16th June.

Watch the Wiki site under activity 3.2.4 for details of the training courses.

Thursday 12 January
8.30 am: JM Ribaut
Jean-Marcel outlined the objectives of the Workshop Presentation.

  1. Review the current status of tools development in Objective 2.2 of the current Proposal
  2. Describe the proposed model and process for integrating these tools into the IB Workbench using the iPlant Cyber Infrastructure and developing stand-alone options for the tools.
  3. Revise the Workplan and responsibilities for Objective 2.2 and 2.3 into a single objective designed to deliver the Configurable Workflow System.

Although the Configurable Breeding Workflow System (CWS) should be ready by mid 2012, there is considerable pressure to show progress over the next six months.
Marker services are already open to users via Chunlin He and some other services like breeding application planning are available. Capacity building activities are underway with improvement of phenotyping sites and training on data management and analysis.
Jean-Marcel described the changes in philosophy with respect to the CWS:

  • ICIS has been dissected and some elements will be reused in IBP.
  • There has been a shift in emphasis from a functional perspective to an operational perspective where tools are grouped into applications aimed at specific breeding activities.
  • IBP will not develop the cyber infrastructure (CI), but will be plugged into the iPlant CI. iPlant will provide the CI and guidance to developers integrating the tools.
  • Some researchers developing command-line tools may not make interface and mock-ups. They will pass on scripts to the 3 GCP engineers based in iPlant.
  • The objective is to deliver a first a minimum version of CWS to users by the end of the year. This will be improved little by little by adding new applications to approach state-of-the art analyses.

9 am: Graham McLaren Presentation and Hector Sanchez Presentation
Graham outlined the process by which the IB Fieldbook had been planned and developed with Hector at CIMMYT. A beta-version of the Field book will be tested by users at the workshop in Nigeria in February.
Delphine asked about help icons for users. Mark proposed quick Mac-style video tutorial (1-2 min). Both options will be explored as the Fieldbook developers.
Hector gave a demonstration of the alpha version of the Fieldbook.

10 am: Jiankang Wang Presentation
Jiankang gave an overview of the existing simulation tools QuLine, QuHybrid and QuMARS. What is currently missing for cross prediction is estimates allele/genotype value. This might be provided with the help of Alain Charcosset's tool for multi-allelic populations.
One issue with the QuGene system is that the tools and executables are free but not the source code which is owned by UQ and constrained byt contracts with some users/developers. Naim indicated that iPlant could integrate the executable but couldn't optimise the computing efficiency without access to the source code.
The best way to pass the Qu tools to iPlant was thought to be to build command line versions of the tools and to develop a GUI that iPlant could use.

11 am: Scott Chapman Presentation
Scott described the work to develop an interface to the QUGene simulation system which provides a good format to describe a breeding program and allows the user to:

  • propose different strategies
  • draw breeding graphs
  • show phenotye and alleles information

There was a discussion of the distinction between tactical and strategic questions which can be addressed by simulation. Mark and Scott know that breeders have downloaded QuGene but don't actually know who use it. Strategic breeding questions are complex and require considerable planning and model preparation. Breeders generally need technical assistance with this sort of question. It is possible to develop applications which use fewer options of the QuGene system to answer tactical questions and these should be targeted for the CWS. Graham suggested we should consider offering strategic simulation as a IBP service, but integrate tactical simulation into other CWS applications. The simulation group agreed to consider and specify the tools which should be integrated into the CWS.

12 pm: Trushar Shah
Trushar described the Cross Predictor tool that calculates breeding values based on alleles for known genes (eg: grain quality of wheat, Howard Eagles). This requires very good data resources with phenotypic values for many known genotypes. Trushar will follow up with Alain to compare and integrate their tools which seem to present some similarity although Trushar 's tool is really a visualisation tool, while Alain's tool identifies best crosses to be made based on a ideal genotypes that might be designed on quite a large number of QTLs . The other approach to Cross Prediction is to use simulation and this might be the preferred approach when less empirical data is available and when the genetic systems are less well
understood. Trushar to consider both approaches when designing a project to develop a Cross Predictor.

2 pm: Naim Matasci (iPlant) Presentation
Naim gave an overview of the way in which command line tools can be integrated into the iPlant cyber infrastructure. The first release of the CI is due this month and then the next one will be in June. Naim gave a Demo of an iPlant workspace where people can do analyses.
Although iPlant resources limited in number of developers, they have software specialists who can help with the integration. Collaborators need to send an email to iPlant with information on the parameters they need for the tool (cf slide). Dialogue is important with the tool developer. iPlant uses JSON-format to specify interfaces and job flows. If an R script is provided this is the same procedure as any other command line application. Options will be turned into a set of arguments. The developer' Script passes as an argument in iPlant script.
However, iPlant has difficulties at the moment to integrate applications with their own GUI, there is no standard process at the moment. For the Field book, iPlant would use Java script from Hector and create a new GUI within the iPlant environment.
All scripts are executed in Arizona or Texas, and this process is invisible to users. Scott asked about remote access without internet or phone. Graham responded that the suggested solution was to provide a package that can run locally, with automatic synchronisation with db. Naim indicated that one could have the database at iPlant or at an independent site.
Naim gave a demonstration of the Single Site Analysis script running on the iPlant infrastructure. Scott observes that it takes too long to run each step of the analyses on the remote system. Users are likely to give up. So we need to bundle the steps or run them locally. We need to consider the option of the iPlant CI detecting a local installation of R and using that when possible.
Graham remarked that to process of integrating scripts by sending JASON scripts to iPlant by email and waiting for them to be mounted there before reviewing and testing was not responsive enough to develop a complex workflow interface. What is needed as an environment where collaborators can mount and modify scripts themselves interactively. Naim was concerned about the possibility of errors endangering the integrity of the running system. Graham suggested that an isolated test environment of an exportable environment to run locally was needed.
There was a discussion about chaining scripts into a workflow. Should output be exported at every step and reimport for the next tool? Graham responded that this was not necessary unless user intervention was required provided a standard for passing information was established. Naim indicated that this should be ok in iPlant.
Scott remarked that he has a tool to draw the plots, make distribution graphs, normality tests etc. It produces 1 page per trait to quickly check for outliers and errors. If users are not happy with the interface, iPlant could modify it accordingly, to show parameters in boxes that the user can tick. A Wizard is probably the best approach.
Jean-Marcel questioned the need for mock-ups for the user interface for all applications. Naim indicated that iPlant uses requirements specialists to define the interface and not necessarily mock ups. The standard process is to transform tools into a box-windows style. Jean-Marcel indicated that there was some risk of loosing great interfaces in iPlant for tools that already exist. Scott mentioned that R has a lot of box-windows packages that could be used. Hector wanted to know what is the real advantage of using iPlant? Naim responded that server maintenance, security maintenance, update of db, configuration and storage resources, community sharing were all supported by iPlant.

3 pm: Shah Trushar Presentation
Trushar gave an overview and demonstration of the molecular breeding decision tool (MBDT) which is capable of the following actions:

  • read genotype, linkage map, QTL file, phenotype files
  • draw chromosomes, showing QTL in colors
  • we can select specific genotypes and choose a diversity panel, sort them by markers, by allele frequency
  • All QTL values can be read in pop-up windows.
  • We can select parents (donor and recurrent)
  • Select target genotypes
  • Can drag/drop marker in background or foreground marker
  • Show lines genotype resulting from cross
  • Use POPMIN to calculate population size necessary

Input and output can be managed with Flapjack format and transformed between formats with ISMAB so that iMAS tools can be used for preparing input to MBDT.
Alain wanted to know if it was possible to see gradient-like heatmaps of the genotypes? Trushar felt that this could be easily done. Fred remarked that breeders need to decide in few weeks before sowing the next generation, so they need systems that provide decision support very quickly. Systems like ISMAB might take too long to be truly used by breeders.
Scott remarked that Flapjack automatically checks for updated versions when it is run. Graham remarked that Eclipse system used to develop MBDT does have this option.
Naim will check how this could be done in iPlant.

4 pm: Fred van Eeuwijk Presentation
Fred gave an overview of the QTL analyses pipeline. Fieldbooks with a full description of the trials: would be extremely useful for the QTL tools and could be used directly. They start with Single Site Analysis scripts have already been written by Jose Crossa's team and also by Biometris. Modeling GxE is useful as quality control for breeding and to answer the question about which environments are similar or different. The the issue of QTLxE can be addressed: is the QTL stable in different environments?
Fred's team use R, but there are still some tools that don't work as well as with GenStat. At this stage there is no user-friendly interface to their R scripts.
See: Detection and use of QTL for complex traits in multiple environments.
van Eeuwijk FA, Bink MC, Chenu K, Chapman SC. Curr Opin Plant Biol. 2010,13:193. Review.
There is a package and book available for R/qtl. One problem is that R/qtl was written for human and animal genetics, so control of local variation and multi-environment analyses steps are missing. Other problems concern the quality of the genetic map. If the genetic map is wrong, there is not much one can do. Biometris will add JoinMap functions to give map construction options to their R tools but only in 2012. Output from their QTL tool is in Flapjack format and so can be passed to other tools easily.
Jeff wanted to know if it could be used by breeders in Africa? Jean-Marcel suggested we should make a tool proposing 2-3 ways to select QTL and genes for the selection step to keep it simple. Fred proposes to create Flapjack output file with visual results and Alain proposes to work on an interpretation of the QTL results aiming to suggest QTLs for using in the next step of selection.
Fred showed the interface to the QTL menus of GenStat. He would like to make an interface to R quite similar, simpler if possible, and orienting users in the analysis. Naim felt this was quite possible. Jean-Marcel felt we should direct the user (even though it looks like patronizing), not showing too many options available, to avoid basic errors.
Fred has a list the R tools ready for integration but the problem is that no one in their group can work on the user interface. Gouyuo said that there is a GUI ready for the R/qtl package that could be used.
One key question was how to help users understand what to do with the list of QTL following the analysis – the decision support step is missing. Jean-Marcel suggested that we should propose 2-3 scenarios to the users. Issues of pleitropy, stability (over environments and backgrounds), traits heritability and correlation with yield and combining effects over loci and traits must be addressed.

5 pm: Alain Charcosset Presentation
Alain gave an overview of the features of OptiMAS a tool for marker assisted recurrent selection. Jeff wanted to know if it is possible to impose a specific QTL to be included in the MARS selected lines. Alain suggested putting maximum weight on this QTL, should work well.
Discussion centered on the same problem as with the standard QTL procedure: how to define favourable/unfavourable alleles across traits, environments and backgrounds. Fred indicated that there is almost nothing published about how to choose target QTL for selection and breeders find it hard to give a satisfying answer.

Conclusions from Day 1:
Jean Marcel indicated that for June, we need to show:

  • the Field book as stand-alone, tested by users
  • the QTL analysis and statistical tools as stand-alone, tested by users
  • a basic pipeline in iPlant that integrates a version of the Fieldbook and the integrated QTL R scripts.

Currently there are limitations with iPlant, especially in terms of integrating graphical user interfaces, but these should be solved in future.
By January 2012 we need integration in iPlant of some modules in the work flow system. We have started recruitment of the 3 software engineers at iPlant. The positions are just opened but when these people are on board we should be able to proceed with the integration in iPlant.
Jeff wanted to know how the use cases would be involved. Jean-Marcel responded that we had identified a group of 3-4 user-cases for each module. To help with the definition of the user interface and test the implementation.

Friday 13 Jan
9 am Naim Matasci, overall presentation of iPlant Presentation
Naim gave an overview of the iPlant Collaborative. The discussion centered on how tools were integrated into the iPlant CI. Naim indicated the following steps:
IBP developers make a mock-up and iPlant will assist developers to integrate their tools. To build on the cyber-infrastructure with iPlant:

  • choose tools
  • describe tools
  • show mock-up
  • data integration: future step at iPlant

The current version of the iPlant CI is a pre-release version. A final version will be released in June. The source code will be released but there are licencing issues at the moment with the University of Arizona. When these are sorted out, then code will be released as open-source.
Alain asked about the users and relationship with iPlant. Naim responded that there was a group of plant experts who meet every fortnight. We could see the Working Group members on the iPlant website: all systems biology experts, no breeders. We are the first crop experts. Jen-Marcel felt that the IBP project brings continuity to their expert groups, downstream with impact and development of public goods. Alain felt that the problem we tackle is specific: genotype-to-phenotype, and quite different to plant systems biology. Jean-Marcel felt we should try to link better with the iPlant Grand challenges and their experts. We should link with other big projects, large communities to get input for implementing IBP, advertise IBP to future users and make it sustainable. Aaron was asked about a big genotype to phenotype project in Wheat that is just starting, and how we could make contact with that group.
Graham remarked that the Department of energy has a similar plan (Kbase) and similar issues to the IBP. The plan to breed energy crops and will use infrastructure compatible with iPlant. They know about our initiative we will try to work together, but it is still not clear how. Jean-Marcel felt we need to be more pro-active with them.

10 am JM Ribaut Presentation
Jean-Marcel reviewed the next steps we must take in remodelling the workplan for Objective 2.2 and 2.3. He indicated that the software engineering positions have been opened, the descriptions are on the IBP website and asked everyone to distribute them to their networks. He indicated that the choice of database was a key issue as apparently there is no community effort to create common database. We could plan another workshop specifically on the database issue. Who should be in this workshop? Not all partners necessary, only key ones, and especially the weaker links. We will also identify a leader for Database.
New members in the Scientific and Management Advisory Committee were suggested:

  • Graeme Hammer (Uni of Queensland) for Crop modelling and prediction tools.
  • Bill Beavis, who is a part of iPlant expert group for Advance use of informatics for modern breeding.

Jean-Marcel described the IBP team assembly:
Management members of IBP/GCP should be much more involved in the coordination of the modules in a flat management structure:

  • Product delivery (A): Larry
  • Community of Practice (E): Ndeye
  • Data management (CIJ): Graham and Arllet
  • Software Development Integration (DHI): Delphine
  • Communication: Antonia
  • Chair: JM

These people are employed by GCP, working partly on IBP in the areas of their specific expertise.
Scott asked about the role of research in the IBP Graham responded that in Objectives 2.2 and 2.3 we need to develop tools for breeders as a first priority. However there was an opportunity for collaborative research with users (especially in simulation for strategic issues) as part of the research services component.
Scott also asked about the environmental data associated with field trials. Jean-Marcel and Graham responded that they should be collected in the Field book. GCP has invested a lot in weather stations, which installed already at several sites. IBP should have a connection with the Climate change Program for centralization of weather and site characterization records.

11 am Graham McLaren
Graham reviewed the team composition for each breeding application of the configurable workflow system (see table in the annexe. Main leaders of each application are in blue.)

  • Application 1. Workbench Administration and Configuration_ (Sanchez_)_Alain asked if we could point out the dependency link between tools. For example, Naming conventions will be used in many other tools. Graham responded that the problem was more about having the schema and tools to manage conventions that each user wanted to specify.
    It was decided to add a sub-application (task) to Application 1 – that of deciding the database back-end and how it will be deployed as stand-alone and in iPlant.
  • Application 2. The Breeding manager{}This application has three sub-applications lead by different people:{}The breeding manager (Sanchez){}The genealogy Manager (Gupyou Ye){}The Query Tools (Chengzhi Liang)_The important thing about the Quesry tools (vis-à-vis the current status) is that they must integrate data across a local and central repository.
  • Application 3. Field Trial management system (Sanchez)_Scott and Mark wanted to know where high throughput electronic data scored in the field or lab will be stored (eg: NIR wavelengths)? Graham responded that summaries could be stored in the fieldbook but we would need new systems for storing the raw data. Questions were asked about storing data measured on bulks and about using barcodes. Graham responded that the sampling tracking facilities of ICIS can handle bulks and that bar-coding was an integral part of the IB Fieldbook.
  • Application 4. Genotypic Data Management System (Trushar Shah)_Graham explained the philosophy of the gene catalogue approach which actually involves a catalogue of molecular variants. It is not know whether this approach is envisaged for the high throughput genotyping management system. It is also not clear how to handle haplotype blocks? Fred remarked that Haplotype Blocks are results from the Analysis tools and that this is a common problem that the private sector (Roche, Keygen, Monsanto...) has already discussed and tackled. He suggested that we shouldn't redo it. Trushar will investigate existing options and adapt them for IBP.
  • Application 5: The Analytical Pipeline (Fred van Eeuwijk)_Fred will be the research leader for the overall application 5. R scripts from different researchers (Fred, Pancho, Trushar, Scott, Guoyou) will be assembled , then passed on to Guoyou for inclusion in a stand-alone menu system based on TCL, and they will be passed on to iPlant for integration in the CI. There is still an issue with the use of REML in R and Fred suggested that Jean-Marcel seek a licence agreement with VSNi for the R version of ASREML.
    • The MARS tool requires a list of QTL targeted for the election program. We need guidelines for breeders to choose target QTL. Fred asserted that we need weights for the traits and the environments. Alain asked what is the allelic effect in a multi-trait context. Jeff indicated that the traits that are important to his program are: drought tolerance in late sowing trials and quality traits. For example in cowpea, grain quality is very important: size, colour, texture. Fred indicated that information on tolerance to specific environment (abiotic stress) will be provided in a correlation table. Fred, Alain and Scott will discuss about decision-making tool for target QTL for selection. We'll need first a quick analytical tool and later an elaborated tool using simulation.
    • Considering Genome Wide Selection, Fred and Delphine want to put it on hold in the IBP as it is still a research tool needing validation in most crops. Alain volunteered to do research on it.
  • Application 6: Decision Support Tools{_}There are four sub-Applications:
    • Breeding applications MBDT, MOSEL and MABC (Trushar)_Graham suggested that the MOSEL specifications migh be easily completed in the MBDT environment.
    • Marker Assisted Recurrent Selection (Alain Charcosset){}Cross Prediction (Trushar Shah){}Simulation Tools (Mark Dieters)_Mark will identify the tactical simulation tools which can be integrated into the workbench

2.30 pm: Jiankang Wang (CAAS)
Jiankang gave a brief overview of how to integrate gene discovery and breeding. The conversation to be integrated with the discussion to be led by Fred, and Jiankang will work with Alain to integrate their respective tools.

3 pm: Jose Crossa (teleconference, CIMMYT) Presentation
Jose gave a presentation via teleconference on the development and use if selection indices for phenotypic and molecular data. All tools are available in SAS and some have been converted to R. Jose showed how to define weights for traits that breeder want or don't want to change. Fred suggested that the method was complicated and proposed to directly use the QTL value instead of going back to the phenotyping value.

4.30 pm Sonya Lowry (teleconference, iPlant)
Sonya joined by teleconference to respond to some questions that had been posed concerning iPlant integration:

  • She confirmed that it was difficult to integrate applications with a desktop GUI right now. She indicated that desktop applications can be integrated in CI as a client. Example: The Field book is a java application, iPlant can replicate it as a web application. There is already a GUI mockup for the Field book and iPlant can replicate the interface. Sonya suggested that a desktop application that is designed to access the shared iPlant CI for some activities should provide the user with a consistent experience whether they connect via the desktop application or the web application.
  • There is no existing program to write JSON script automatically from R scripts. It is on the iPlant near term roadmap to create an application that generates JSON, but it won't do it by parsing R.
  • Hector wants to see the web application and see how the interface will look. Sonya will send it to Hector, when available. Hector will send to Sonia the prototype of the Field book with description and current interface.
  • Steve Goff indicated that iPlant will store sequencing data for different crops (in collaboration with Cornell University). A small group of experts was identifying the strategy to do it.
  • Scott wanted to know if simulation jobs could be submitted to a condor based cluster of widows machines via iPlant. Sonya responded that another virtual machine clustering option (Atmosphere) was available at iPlant.
  • Alain wanted to know how are we going to interact with iPlant experts on other projects? Sonya replied that there is an annual conference of iPlant. Steve indicated that this year, iPlant will organize a workshop for all partners to directly address scientific questions by integrating applications into the CI.
  • The question was asked about which tools iPlant is working on with their partners? Could they list them to avoid redundancy and join force between different projects on similar tools? Steve agrees and would like to publish for each tool, what it does, who works on it, at what stage it is at. Sonya confirmed that this information is being compiled for distribution.

Conclusions from Day 2

  • Jean-Marcel stressed that the objective is to have parts of our CWS operating quite soon.
  • Until March, it won't be possible to translate our script in JSON ourselves. So we need some help from iPlant with this integration.
  • Fred proposed to send some scripts to iPlant for them to see how it works and what it will look like. In the meantime, we improve our scripts to make them more robust. Naim agreed on this strategy.
  • Steve stressed that the more complex it gets, the sooner IGCP should get their software engineers in place. iPlant won't be able to allocate their staff on this.
  • In the meantime, Jean-Marcel proposed that we focus on simple modules: in March, Fred will deliver variance analysis scripts to iPlant. Naim will send samples of JSON to IBP (Scott, Fred) and we will do it.
  • Naim stressed that they won't do the documentation. We have to prepare it.
  • Steve proposed a quick solution: we can contract the work until GCP engineers are hired.
  • Signature with iPlant will be postponed to the iPlant workshop at PAG on Monday evening.
Adaptavist Theme Builder (3.3.3-conf210) Powered by Atlassian Confluence 2.10.3, the Enterprise Wiki.
Free theme builder license