Hey Vincent,
Thanks for your question! I didn’t think about demo-ing the drag and drop feature, my apologies.
So the idea behind bias reduction is that engaging with data that isn’t your own (i.e. you acting as a 3rd party curating a researcher’s data) will mean you create specific inferences and hypotheses for the data from your own knowledge; which may not have been the intent of the researcher who’s data is being presented.
While this can be a good thing, inferences by a 3rd party curator can also end up confusing the researcher and not providing the results they may have desired. By creating a tool in which researchers can lay out the SVGs the way they see fit, and input specific data in which they’re able to make inferences out of the ‘bias’ of a 3rd party is eliminated.
Furthermore, the current layout of the eFPs we provide may not be what the researcher requires, and may introduce a bias to the data; as only certain parts of the data that an eFP depicts are portrayed, rather than the data that the researcher may require.
Hey Wen Kai,
thanks for your response! Unfortunately I just ran out of time with this project, as I was stuck on another section of it and could not get to that aspect of it.
Thank you for your response! Apologies, I don’t think I explained myself well, it wouldn’t be an algorithm that reduces curator bias, but instead the person interacting with the project that causes bias reduction. The idea being that when someone else works with your data, they bring in their own biases in depicting it; and my creating a view that is not what you intended, and so shows a different result. While this isn’t wrong, a researcher may want to highlight certain aspects of their data which they aren’t able to if someone else is creating the visualisation.
I see that the user can pick SVGs representing different tissues, but what about experimental conditions? I assume there is one BAM file per condition, which the user would somehow associate with the selected tissues, so if the researcher has multiple experimental conditions they would have to create multiple images (or is there a way to automate this?)
Thanks for your question! You are correct, it would be one BAM file per condition. So the idea would be to add in an SVG per condition, so you could highlight how different experimental conditions affect expression; and the drag-and-drop functionality allows cloning for that reason. You would be able to add multiple copies of the same SVG on the single image you create, so that you can visualise these differences. Currently there is no way to automate this, but that could be a really interesting future direction.
Good job! Your presentation was very thoughtful and the scope of the project was clear and concise. For my question, I was wondering if there is any sort of quality control was done on the reads for the BAM files on the back-end.
So you mention that this tool will “reduce bias of a curator”. In what ways specifically?
We would appreciate us showing you dragging those ‘icons’ onto the canvas – that took some effort!
Hey Vincent,
Thanks for your question! I didn’t think about demo-ing the drag and drop feature, my apologies.
So the idea behind bias reduction is that engaging with data that isn’t your own (i.e. you acting as a 3rd party curating a researcher’s data) will mean you create specific inferences and hypotheses for the data from your own knowledge; which may not have been the intent of the researcher who’s data is being presented.
While this can be a good thing, inferences by a 3rd party curator can also end up confusing the researcher and not providing the results they may have desired. By creating a tool in which researchers can lay out the SVGs the way they see fit, and input specific data in which they’re able to make inferences out of the ‘bias’ of a 3rd party is eliminated.
Furthermore, the current layout of the eFPs we provide may not be what the researcher requires, and may introduce a bias to the data; as only certain parts of the data that an eFP depicts are portrayed, rather than the data that the researcher may require.
Nice explanation. what is the challenge to merge the icons and the statistic data generated?
Hey Wen Kai,
thanks for your response! Unfortunately I just ran out of time with this project, as I was stuck on another section of it and could not get to that aspect of it.
I think it’s a useful project. But I am curious about the algorithm of reducing the bias of curator, could you please talk a little bit about that?
Hey Yue Xin,
Thank you for your response! Apologies, I don’t think I explained myself well, it wouldn’t be an algorithm that reduces curator bias, but instead the person interacting with the project that causes bias reduction. The idea being that when someone else works with your data, they bring in their own biases in depicting it; and my creating a view that is not what you intended, and so shows a different result. While this isn’t wrong, a researcher may want to highlight certain aspects of their data which they aren’t able to if someone else is creating the visualisation.
Hi Sakshi,
I see that the user can pick SVGs representing different tissues, but what about experimental conditions? I assume there is one BAM file per condition, which the user would somehow associate with the selected tissues, so if the researcher has multiple experimental conditions they would have to create multiple images (or is there a way to automate this?)
Hey Bruno!
Thanks for your question! You are correct, it would be one BAM file per condition. So the idea would be to add in an SVG per condition, so you could highlight how different experimental conditions affect expression; and the drag-and-drop functionality allows cloning for that reason. You would be able to add multiple copies of the same SVG on the single image you create, so that you can visualise these differences. Currently there is no way to automate this, but that could be a really interesting future direction.
Hi Sakshi,
Good job! Your presentation was very thoughtful and the scope of the project was clear and concise. For my question, I was wondering if there is any sort of quality control was done on the reads for the BAM files on the back-end.
Cheers,
Rachel