Give a first pass at timelines for project works

Oct 13, 2025 by magicalceylontours.com 28 views

Introduction

In-the-loop modelling is a powerful approach to machine learning that involves human feedback and interaction to improve model performance and accuracy. However, this approach requires a significant amount of high-quality data to train and validate the models. In this article, we will discuss the importance of data generation for in-the-loop modelling and outline a plan of work to achieve this goal.

The Importance of Data Generation

Data generation is a critical step in the development of in-the-loop models. The quality and quantity of the data used to train the model can significantly impact its performance and accuracy. In-the-loop models require a large amount of data to learn from, and the data must be annotated with relevant information such as cell line, cell type, tissue, and synthetic tissue. The data must also be diverse and representative of the real-world scenarios that the model will be applied to.

Plan of Work

Following discussion with Santosh, @amina-mardiyyah & @withers-ebi are keen to follow the plan of work laid out by Melanie Vollmar in Nature paper. The plan of work involves several key steps:

Step 1: Annotating Papers

Discuss the need for in-house assistance gathering ~10 annotated papers: The first step is to discuss the need for in-house assistance gathering ~10 annotated papers with Barbara and Melissa. This will involve identifying the relevant papers and annotating them with the required information such as cell line, cell type, tissue, and synthetic tissue.
Use ChEMBL assay descriptions to select more ‘known’ cell lines: Christine will use the ChEMBL assay descriptions to select more ‘known’ cell lines. From these search in literature for these ~10 papers to annotate.
Review using LabelBox: Christine will keep an eye on Leadmine licensing and review using LabelBox.

Step 2: Preparing the Model Pipeline

Prepare showing how we are ready-to-go with a packaged up model pipeline: Amina and Christine will prepare showing how we are ready-to-go with a packaged up model pipeline that is already fit for human-in-the-loop.
Give a first pass at timelines for project works: Amina and Christine will give a first pass at timelines for project works and maybe refine our model pipeline at Miro.

Conclusion

In-the-loop modelling is a powerful approach to machine learning that requires high-quality data to train and validate the models. Data generation is a critical step in the development of in-the-loop models, and it requires a significant amount of annotated data to learn from. The plan of work outlined in this article involves several key steps, including annotating papers, preparing the model pipeline, and giving a first pass at timelines for project works. By following this plan of work, we can ensure that we have a comprehensive approach to data generation for in-the-loop modelling.

Future Work

Future work will involve refining the model pipeline, testing the model on a larger dataset, and evaluating the model's performance and accuracy. We will also need to consider the scalability and deployment of the model in a real-world setting.

References

Nature paper

Action Items

[ ] Discuss w Barbara and Melissa the need for in-house assistance gathering ~10 annotated papers (annotating for cell line, cell type, tissue, synthetic tissue)
[ ] Christine to use the ChEMBL assay descriptions to select more ‘known’ cell lines - from these search in literature for these ~10 papers to annotate
[ ] Christine to keep an eye on Leadmine licensing, and to review using LabelBox
[ ] Amina & Christine to prepare showing how we are ready-to-go with a packaged up model pipeline that is already fit for human-in-the-loop
[ ] Amina & Christine to give a first pass at timelines for project works & maybe refine our model pipeline at Miro
Data Generation for In-the-Loop Modelling: A Q&A Guide ===========================================================

Introduction

In-the-loop modelling is a powerful approach to machine learning that involves human feedback and interaction to improve model performance and accuracy. However, this approach requires a significant amount of high-quality data to train and validate the models. In this article, we will answer some of the most frequently asked questions about data generation for in-the-loop modelling.

Q&A

Q: What is in-the-loop modelling?

A: In-the-loop modelling is a type of machine learning approach that involves human feedback and interaction to improve model performance and accuracy. This approach requires a significant amount of high-quality data to train and validate the models.

Q: Why is data generation important for in-the-loop modelling?

A: Data generation is a critical step in the development of in-the-loop models. The quality and quantity of the data used to train the model can significantly impact its performance and accuracy. In-the-loop models require a large amount of data to learn from, and the data must be annotated with relevant information such as cell line, cell type, tissue, and synthetic tissue.

Q: What are the key steps involved in data generation for in-the-loop modelling?

A: The key steps involved in data generation for in-the-loop modelling include annotating papers, preparing the model pipeline, and giving a first pass at timelines for project works. These steps are outlined in the plan of work outlined in this article.

Q: How do I annotate papers for in-the-loop modelling?

A: Annotating papers for in-the-loop modelling involves identifying the relevant papers and annotating them with the required information such as cell line, cell type, tissue, and synthetic tissue. This can be done using tools such as LabelBox or by manually annotating the papers.

Q: What is the role of ChEMBL assay descriptions in data generation for in-the-loop modelling?

A: ChEMBL assay descriptions are used to select more ‘known’ cell lines. From these search in literature for these ~10 papers to annotate. This helps to ensure that the data used to train the model is diverse and representative of the real-world scenarios that the model will be applied to.

Q: How do I prepare the model pipeline for in-the-loop modelling?

A: Preparing the model pipeline for in-the-loop modelling involves packaging up the model pipeline that is already fit for human-in-the-loop. This involves ensuring that the model is trained on a large amount of high-quality data and that it is able to learn from human feedback and interaction.

Q: What are the benefits of in-the-loop modelling?

A: The benefits of in-the-loop modelling include improved model performance and accuracy, increased efficiency, and reduced costs. In-the-loop modelling also allows for the development of more complex and nuanced models that can learn from human feedback and interaction.

Conclusion

In-the-loop modelling is a powerful approach to machine learning that requires high-quality data to train and validate the models. Data generation is a critical step in the development of in-the-loop models, and it requires a significant amount of annotated data to learn from. By following the plan of work outlined in this article and answering the frequently asked questions about data generation for in-the-loop modelling, we can ensure that we have a comprehensive approach to data generation for in-the-loop modelling.

Future Work

References

Nature paper

Action Items

[ ] Discuss w Barbara and Melissa the need for in-house assistance gathering ~10 annotated papers (annotating for cell line, cell type, tissue, synthetic tissue)
[ ] Christine to use the ChEMBL assay descriptions to select more ‘known’ cell lines - from these search in literature for these ~10 papers to annotate
[ ] Christine to keep an eye on Leadmine licensing, and to review using LabelBox
[ ] Amina & Christine to prepare showing how we are ready-to-go with a packaged up model pipeline that is already fit for human-in-the-loop
[ ] Amina & Christine to give a first pass at timelines for project works & maybe refine our model pipeline at Miro