![\"dataframe-showing-dupicates\"](\"https://i.ibb.co/fV3VNjR/Screen-Shot-2021-02-08-at-7-51-31-AM.png\")
Hi there.
\nIt appears that 146 of the images contain duplicate classes (just one label duplicated) in their labels. This duplication only happens with the 9, 13, & 17 labels (Actin Filaments, Plasma Membrane, & Vesicles)
I was hoping to understand if this is just a mistake or if it indicates something special?
\nI have attached a link to the dataset containing the CSV shown above.
\n", "messages": [ { "id": 1191385, "postDate": "2021-02-08T12:57:53.410Z", "content": "Hi there.
\nIt appears that 146 of the images contain duplicate classes (just one label duplicated) in their labels. This duplication only happens with the 9, 13, & 17 labels (Actin Filaments, Plasma Membrane, & Vesicles)
I was hoping to understand if this is just a mistake or if it indicates something special?
\nI have attached a link to the dataset containing the CSV shown above.
\n", "rawMarkdown": "Hi there.\n\nIt appears that 146 of the images contain duplicate classes (just one label duplicated) in their labels. This duplication only happens with the **`9, 13, & 17`** labels (**`Actin Filaments, Plasma Membrane, & Vesicles`**)\n\n---\n\n\n\n---\nI was hoping to understand if this is **just a mistake or if it indicates something special?** \n\nI have attached a [link to the dataset](https://www.kaggle.com/dschettler8845/hpa-duplicate-classes-dataframe) containing the CSV shown above.\n\n- https://www.kaggle.com/dschettler8845/hpa-duplicate-classes-dataframe",
"votes": 27
},
{
"id": 1193562,
"postDate": "2021-02-09T17:55:14.597Z",
"content": "Hi! In this competition, we merged some spatially similar (and functionally related) patterns into the same class.
\nThis was mentioned on the Single Cell Pattern notebook
\nThe duplicates simply results from conversion code. For example, if an image has image-level label as Vesicles and Peroxisomes, then the code results in 17|17. In specific for the labels you mention above:
9. Actin filaments\nThis class includes Actin filaments and Focal adhesion sites.\n13. Plasma membrane\nThis class includes plasma membrane and cell junctions\n17. Vesicles and punctate cytosolic patterns\nThis class includes small circular compartments in the cytosol: Vesicles, Peroxisomes (lipid metabolism), Endosomes (sorting compartments), Lysosomes (degradation of molecules or eating up dead molecules), Lipid droplets (fat storage), Cytoplasmic bodies (distinct granules in the cytosol).\nAlso to point out the last merge that didn't result in 'duplicate'
\n12. Centrosome\nThis class includes centrosomes and centriolar satellites.\nPerfect. Thank you for clearing this up.
", "rawMarkdown": "Perfect. Thank you for clearing this up.", "votes": 1 }, { "id": 1237050, "postDate": "2021-03-13T17:21:44.890Z", "content": "Sorry if I've missed something, but could you please also comment on a few other patterns from the previous HPA competition? Specifically: Rods & Rings, Microtubule ends, Microtubule organizing center, and Cytokinetic bridge. Could you please tell us how there were combined with others, or whether they're just not present?
", "rawMarkdown": "Sorry if I've missed something, but could you please also comment on a few other patterns from the previous HPA competition? Specifically: Rods & Rings, Microtubule ends, Microtubule organizing center, and Cytokinetic bridge. Could you please tell us how there were combined with others, or whether they're just not present?" }, { "id": 1237508, "postDate": "2021-03-14T07:58:07.830Z", "content": "Hi! Other classes are removed in this competition because of considerations for single cells. For example Cytokinetic bridge are best observed between 2 cells and unreliable if looking just at 1 cell.
Thanks for sharing this (excellent observation!), might be it indicates multiple elements in the cell (some sort of mutation or disease). Competition organizers can best answer this question.
Hi! In this competition, we merged some spatially similar (and functionally related) patterns into the same class.
\nThis was mentioned on the Single Cell Pattern notebook
\nThe duplicates simply results from conversion code. For example, if an image has image-level label as Vesicles and Peroxisomes, then the code results in 17|17. In specific for the labels you mention above:
9. Actin filaments\nThis class includes Actin filaments and Focal adhesion sites.\n13. Plasma membrane\nThis class includes plasma membrane and cell junctions\n17. Vesicles and punctate cytosolic patterns\nThis class includes small circular compartments in the cytosol: Vesicles, Peroxisomes (lipid metabolism), Endosomes (sorting compartments), Lysosomes (degradation of molecules or eating up dead molecules), Lipid droplets (fat storage), Cytoplasmic bodies (distinct granules in the cytosol).\nAlso to point out the last merge that didn't result in 'duplicate'
\n12. Centrosome\nThis class includes centrosomes and centriolar satellites.\nPerfect. Thank you for clearing this up.
", "votes": 1, "replies": [] }, { "id": 1237050, "author_name": "V. Gates", "author_url": "", "post_date": "2021-03-13T17:21:44.890000", "content": "Sorry if I've missed something, but could you please also comment on a few other patterns from the previous HPA competition? Specifically: Rods & Rings, Microtubule ends, Microtubule organizing center, and Cytokinetic bridge. Could you please tell us how there were combined with others, or whether they're just not present?
", "votes": 0, "replies": [] }, { "id": 1237508, "author_name": "Trang Le", "author_url": "", "post_date": "2021-03-14T07:58:07.830000", "content": "Hi! Other classes are removed in this competition because of considerations for single cells. For example Cytokinetic bridge are best observed between 2 cells and unreliable if looking just at 1 cell.
Thanks for sharing this (excellent observation!), might be it indicates multiple elements in the cell (some sort of mutation or disease). Competition organizers can best answer this question.
\n\n---\nI was hoping to understand if this is **just a mistake or if it indicates something special?** \n\nI have attached a [link to the dataset](https://www.kaggle.com/dschettler8845/hpa-duplicate-classes-dataframe) containing the CSV shown above.\n\n- https://www.kaggle.com/dschettler8845/hpa-duplicate-classes-dataframe",
"1193562": "Hi! In this competition, we merged some spatially similar (and functionally related) patterns into the same class.\nThis was mentioned on the [Single Cell Pattern notebook](https://www.kaggle.com/lnhtrang/single-cell-patterns)\nThe duplicates simply results from conversion code. For example, if an image has image-level label as Vesicles and Peroxisomes, then the code results in `17|17`. In specific for the labels you mention above:\n\n```\n9. Actin filaments\nThis class includes Actin filaments and Focal adhesion sites.\n```\n\n```\n13. Plasma membrane\nThis class includes plasma membrane and cell junctions\n```\n\n```\n17. Vesicles and punctate cytosolic patterns\nThis class includes small circular compartments in the cytosol: Vesicles, Peroxisomes (lipid metabolism), Endosomes (sorting compartments), Lysosomes (degradation of molecules or eating up dead molecules), Lipid droplets (fat storage), Cytoplasmic bodies (distinct granules in the cytosol).\n```\n\nAlso to point out the last merge that didn't result in 'duplicate'\n```\n12. Centrosome\nThis class includes centrosomes and centriolar satellites.\n```",
"1192467": "Thanks for sharing this (excellent observation!), might be it indicates multiple elements in the cell (some sort of mutation or disease). Competition organizers can best answer this question."
}
}