# Memo for management (DRAFT)
## Refocus of VIB Bioinformatics Core Software Development Service Pillar

## Background
Computational science (science relying on computing) is the new norm, so much so that it is often unrecognisable from even a decade ago. Software and data underpin these achievements. And, while data is regarded with great importance, software often takes a back seat.

A lot of _ad-hoc_, special purpose, single use software is produced by VIB research groups by researchers and bioinformaticians who often lack the specialised skills to create high-quality, general purpose software.

Due the the short-sighted nature of this software development, little or no time and effort is dedicated to the important tasks of documentation, quality control through testing, and packaging for wider use. Even worse, a lot of time, effort, and money is wasted in redundantly developing solutions to problems that have already been solved over, and over again.

Very often, this _ad-hoc_ software is published to the outside world as part of a research publication. Most of the time, users accept the result as just another piece of "research quality" software: VIB and VIB researchers are no better than anyone else in producing software. The worst-case scenario is that bugs in this software result in the retraction of the paper. It is only a matter of time before this happens to VIB as it has happened to so many other institutions.

Furthermore, when the authors of this software move on to other positions (as often happens with PhD students and post-doctoral researchers) or they reach the limit of their time or skill required to improve the software, it falls upon VIB Bioinformatics Core to rescue the project. Clearly, the volume of requests for this type of assistance cannot be matched by VIB Bioinformatics Core.

The software developers employed by the the VIB Bioinformatics Core have the skills and experience required, not only to rescue and maintain this existing software, but also to develop new software to a very high standard of quality in a shorter period of time than non-professional software developers. Not only this, but the VIB Bioinformatics Core has an excellent track record (see Annex) of software projects and an high-level overview of existing research requirements allowing us to produce general purpose software that has a broad impact.

## Proposed refocus of VIB Bioinformatics Core Software Development efforts
The rising tide for software repair and maintenance work shows no sign of slowing. Rather than attempting to meet this demand, the VIB Bioinformatics Core should be focused on preventing this situation. This can be achieved by _initiating_ the development of good, general purpose, and useful research software to a _product quality level_ in partnership with research groups. This does not mean forgoing the valuable role we currently play in improving the quality of existing software produced by research labs, even up to product quality.

Therefore the **two primary priorities** for software development by the VIB Bioinformatics core as proposed are:
1. Focus software engineering effort onto a small number of very high impact projects that are of broad utility to all VIB researchers.
2. Continue to maintain and improve the high-quality existing software the VIB Bioinformatics Core has already worked on (see Annex), and also to support existing software produced by research labs.

The software development partnership between the VIB Bioinformatics Core and research groups can be extended to VIB HQ IT Services who can integrate their efforts with _VIB Services_ including for payment tools, authorisation, and user experience design.

VIB Bioinformatics Core faces two main challenges in achieving the proposed refocus. First, VIB researchers will still require narrow, special purpose software. So the VIB Bioinformatics Core will need to select projects carefully based on criteria such as broad utility, impact on VIB research interests, and on global exposure for VIB as a quality software "brand". VIB Bioinformatics Core is well placed to recognise important projects that meet these criteria in consultation with scientific experts.

The second challenge is that of image and branding in the minds of the researchers and the bioinformaticians who will be our partners. VIB Bioinformatics Core is not _currently_ well known as a centre for software development (despite the success already demonstrated in delivering high quality software projects; see Annex). This perception will need to change if the VIB Bioinformatics Core is going to partner with research groups to produce widely used, product quality software. Meeting this challenge will require assistance and advice from resources external to the VIB Bioinformatics Core. For example, the VIB Bioimaging Core went through a similar image transformation two years ago. Thus, such a transformation is possible, but VIB Bioinformatics Core can also learn from their experiences and use tactics that worked for them.

If the above challenges are successfully overcome, the VIB Bioinformatics Core will be positioned as an _accelerator_ for innovation within VIB. There are a number of significant advantages for VIB as a whole. These include:
1. Maximise opportunities for business development built on product quality software.
2. Less waste in producing redundant software solutions, increasing scientific productivity.
3. A reputation among scientists and businesses the world over for VIB producing high quality software.
4. Reducing the risk of disaster retractions.
5. A tight and efficient training loop provided by VIB Bioinformatics Core training offering working closely with software developers (providing direct feedback from users).
6. Increased visibility from quality software engineering attracts graduates and experienced engineers from industry.

## Potential Use Cases
There are already several potential projects with the broad impact and utility for VIB researchers that may meet the selection criteria detailed above. These projects have not yet been attempted due the the significant time and development resources they would require:
* A combined multi-omics database and software portal seamlessly integrating visualisation and analysis tools in one place. (Backers: Bart Ghesquiere)
* Deeplearning / IONBot business development project (backers: Lennart Martens, 1 year)
* ... [link to grand challenges].
* Deeplearning / web-based collaborative image annotation and automated segmentation? (backers: VIB Bioimaging Core, Yvan Saeys?)
* spatial transcriptomics / Akoya ? (backers: TechWatch / Bart De Strooper /...?)

## Software Product Development Team

The software development team would consist of 5 software engineers
- 2 FTE Software developer (2 new recruits)
- 1 FTE Software developer (James Collier)
- 1 FTE Software developer (Thomas van Parys)
- 1 FTE Senior software developer (Frank Vernaillen)

These engineers would primarily focus on software product development (the first priority as detailed earlier in this document) but could be lent, for up to 1 year, to research labs to improve existing software (the second priority as detailed earlier in this document).

## Request to Management

* 2 FTE (70 kEUR) Software Developers.
* 25 kEUR resources outsourcing maintenance.
* 5 kEUR / yr for consumables such as web hosting.

Where absolutely necessary, maintenance and rescue of existing projects that are of a lower priority may be outsourced to commercial development companies.

## Annex

_VIB-UGent Center for Inflammation Research_
• ConTra V3: an update of the online tool for the exploration of predicted transcription factor binding sites - http://bioit2.irc.ugent.be/contra/v3/
The article on ConTra 3 has been published in the 2017 Web Server edition of Nucleic Acids Research: https://dx.doi.org/10.1093/nar/gkx376 (Saeys lab)

_VIB Bioimaging Core_
• An Interactive ImageJ Plugin for Semi-Automated Image Denoising in Electron Microscopy
https://doi.org/10.1038/s41467-020-14529-0

_VIB-UGent Center for Plant System Biology_
• PhyD3: a new online tool for phylogenetic tree visualization - https://phyd3.bits.vib.be
This tool has been integrated in the PLAZA framework
(http://bioinformatics.psb.ugent.be/plaza/), and published in the journal of Bioinformatics: http://dx.doi.org/10.1093/bioinformatics/btx324
The preprint version is available here: https://doi.org/10.1101/107276

• WgDotPlot: a plugin for visualization of colinear regions between and/or within species
https://bioinformatics.psb.ugent.be/plaza/versions/plaza_v3_dicots/dotplot/index
This tool has been integrated in the PLAZA framework
(https://bioinformatics.psb.ugent.be/plaza/) and published in the Nucleic Acids
Research https://doi.org/10.1093/nar/gkx1002

• TriForC: an online database for high-value triterpenes https://bioinformatics.psb.ugent.be/triforc/
This tool has been published in the database issue of Nucleic Acids Research:
https://doi.org/10.1093/nar/gkx925 (Goossens Lab)

• Plant PTM Viewer: an intuitive protein-centric PTM viewer (De Smet Lab) https://www.psb.ugent.be/webtools/ptm-viewer/
https://doi.org/10.1111/tpj.14345

_VIB-UGent Center for Medical Biotechnology_
• Scop3D 2.0 – online visualization of mutation rates on protein structure (Martens Lab) https://pubs.acs.org/doi/10.1021/acs.jproteome.8b00681

_VIB-KU Leuven Center for Microbiology_
• Mobile application for Vlaams Darmflora project on Android devices, for participants data gathering.
Inhouse development. Screenshot see: https://www.eoswetenschap.eu/gezondheid/dit-wordt-wereldwijd-de-ultiemereferentie-voor-onderzoek-naar-darmflora (in Dutch) (Raes Lab)

• Beerome – Verstrepen Lab https://beerome-rgxf2dt3na-ew.a.run.app/

• CAMEL-DB – Michiels Lab https://cameldatabase.com

_VIB-KU Leuven Center for Brain and Disease Research_
• SCope: An online tool to explore and compare single-cell datasets across species.
This tool has been published in the article: A Single-Cell Transcriptome Atlas of the Aging Drosophila Brain. Cell, 2018 https://doi.org/10.1016/j.cell.2018.05.057
The preprint version is available here: https://doi.org/10.1101/237420 (Aerts Lab).

• AutoSpill (Liston Lab) https://autospill.vib.be
Preprint version: https://doi.org/10.1101/2020.06.29.177196

_VIB-VUB Center for Structural Biology_
• iRefIndex - (Wodak Lab): an update of consolidated protein interaction data and their supporting evidence - https://irefindex.vib.be
It is also available via the PSICQUIC Viewer at EBI: PSICQUIC Viewer