\n
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F190338%2F93c3c302cb5ba4d97ff3f9a65e9b7b23%2Fscreenshot.PNG?generation=1664466911916646&alt=media\")
2022-09-10: The rescore is now complete. I'll check on Monday (09-12) that all previously successful submissions are still successful.
\nTLDR: There is some bad data the public test set (CITEseq day 2 / donor 27678). You should still make predictions for these cells, but they won't affect your score. We are adding the corrected data to a new file.
\nThere was a problem with certain instances of the public test set having been duplicated from the training set. Specifically, what were supposed to be the CITEseq day 2 / donor 27678 cells (public test set) were actually cells for day 2 / donor 32606 (training set).
\nTo avoid creating conflicts with current code and submissions, we'll be keeping the original data the same but are updating the dataset to include the inputs for CITEseq day 2 / donor 27678 in a separate file, test_cite_inputs_day_2_donor_27678.h5 with metadata in metadata_cite_day_2_donor_27678.csv. You should not submit predictions for these new cells, however. (The evaluation_ids.csv are unchanged.)
We're also rescoring all submissions to date with the predictions from the duplicated cells ignored. All submissions going forward will also have these predictions ignored. You should still submit predictions for these cells, but they will not affect your score. (So just a constant 0.0 is fine.)
Please let us know if you have any questions or concerns!
", "messages": [ { "id": 1930287, "postDate": "2022-09-07T17:42:08.723Z", "content": "2022-09-10: The rescore is now complete. I'll check on Monday (09-12) that all previously successful submissions are still successful.
\nTLDR: There is some bad data the public test set (CITEseq day 2 / donor 27678). You should still make predictions for these cells, but they won't affect your score. We are adding the corrected data to a new file.
\nThere was a problem with certain instances of the public test set having been duplicated from the training set. Specifically, what were supposed to be the CITEseq day 2 / donor 27678 cells (public test set) were actually cells for day 2 / donor 32606 (training set).
\nTo avoid creating conflicts with current code and submissions, we'll be keeping the original data the same but are updating the dataset to include the inputs for CITEseq day 2 / donor 27678 in a separate file, test_cite_inputs_day_2_donor_27678.h5 with metadata in metadata_cite_day_2_donor_27678.csv. You should not submit predictions for these new cells, however. (The evaluation_ids.csv are unchanged.)
We're also rescoring all submissions to date with the predictions from the duplicated cells ignored. All submissions going forward will also have these predictions ignored. You should still submit predictions for these cells, but they will not affect your score. (So just a constant 0.0 is fine.)
Please let us know if you have any questions or concerns!
", "rawMarkdown": "*2022-09-10: The rescore is now complete. I'll check on Monday (09-12) that all previously successful submissions are still successful.*\n\n**TLDR: There is some bad data the public test set (CITEseq day 2 / donor 27678). You should still make predictions for these cells, but they won't affect your score. We are adding the corrected data to a new file.**\n\nThere was a problem with certain instances of the public test set having been duplicated from the training set. Specifically, what were supposed to be the CITEseq day 2 / donor 27678 cells (public test set) were actually cells for day 2 / donor 32606 (training set).\n\nTo avoid creating conflicts with current code and submissions, we'll be keeping the original data the same but are updating the dataset to include the inputs for CITEseq day 2 / donor 27678 in a separate file, `test_cite_inputs_day_2_donor_27678.h5` with metadata in `metadata_cite_day_2_donor_27678.csv`. You should not submit predictions for these new cells, however. (The `evaluation_ids.csv` are unchanged.)\n\nWe're also rescoring all submissions to date with the predictions from the duplicated cells ignored. All submissions going forward will also have these predictions ignored. You should still submit predictions for these cells, but they will not affect your score. (So just a constant `0.0` is fine.)\n\nPlease let us know if you have any questions or concerns!", "votes": 50 }, { "id": 1930484, "postDate": "2022-09-07T23:07:46.807Z", "content": "Thank you for this update. I have a question.
\nWhy are all my files making errors after rescore? We don't need to do anything according to your explanation.
The same, 14 submissions from 23 were scored as an error (internal score error code).
", "rawMarkdown": "The same, 14 submissions from 23 were scored as an error (internal score error code)." }, { "id": 1930490, "postDate": "2022-09-07T23:22:58.733Z", "content": "The same, 6 submissions from 11were scored as an error .Is there a maximum of 5 re-scoring?
", "rawMarkdown": "The same, 6 submissions from 11were scored as an error .Is there a maximum of 5 re-scoring?" }, { "id": 1930492, "postDate": "2022-09-07T23:30:52.147Z", "content": "Same issue here, I don't care to count how many of ours have errored, needless to say it's the majority of them.
", "rawMarkdown": "Same issue here, I don't care to count how many of ours have errored, needless to say it's the majority of them." }, { "id": 1930494, "postDate": "2022-09-07T23:54:29.410Z", "content": "Almost 100 participants including me were gone. I think it's same issue as mine.
", "rawMarkdown": "Almost 100 participants including me were gone. I think it's same issue as mine.", "votes": 1 }, { "id": 1930687, "postDate": "2022-09-08T06:04:58.660Z", "content": "Too many submissions get errors. This is non-code competition, imagine what will happen if this is a ode-competition. This makes me wonder if previous code competitions had similar issues as this one or not.
", "rawMarkdown": "Too many submissions get errors. This is non-code competition, imagine what will happen if this is a ode-competition. This makes me wonder if previous code competitions had similar issues as this one or not." }, { "id": 1930760, "postDate": "2022-09-08T07:07:59.313Z", "content": "It is possible some will show as Error until the rescoring is complete. Still have some error but some have scores and all are versions of the same notebook. Have seen this before in other competitions that had rescoring happen, so hopefully all will be OK when it is finished.
", "rawMarkdown": "It is possible some will show as Error until the rescoring is complete. Still have some error but some have scores and all are versions of the same notebook. Have seen this before in other competitions that had rescoring happen, so hopefully all will be OK when it is finished. " }, { "id": 1930797, "postDate": "2022-09-08T07:50:37.777Z", "content": "I think the rescoring is still in progress. Some of my submissions show Error this morning and turn to score later. I also noticed that the top 1 climbed from 6-7th to 1st place without any submission last 10 hours (he was 1st before the rescoring, drop to 6-7th and then return to his place)
", "rawMarkdown": "I think the rescoring is still in progress. Some of my submissions show Error this morning and turn to score later. I also noticed that the top 1 climbed from 6-7th to 1st place without any submission last 10 hours (he was 1st before the rescoring, drop to 6-7th and then return to his place)" }, { "id": 1931090, "postDate": "2022-09-08T12:59:33.193Z", "content": "Apologies for the confusion. It looks like the rescore failed about halfway through. (It may take awhile to work through all the submissions.) I'll run it again and alert our engineering team.
", "rawMarkdown": "Apologies for the confusion. It looks like the rescore failed about halfway through. (It may take awhile to work through all the submissions.) I'll run it again and alert our engineering team.", "votes": 5 }, { "id": 1931112, "postDate": "2022-09-08T13:05:45.227Z", "content": "I have only 3 submissions scored sucessfully. All others got errors.
", "rawMarkdown": "I have only 3 submissions scored sucessfully. All others got errors.", "replies": [ { "id": 1931142, "postDate": "2022-09-08T13:27:59.283Z", "content": "We'll make sure that all submissions that were successful before remain successful. I'm working on getting the rescore started again.
", "rawMarkdown": "We'll make sure that all submissions that were successful before remain successful. I'm working on getting the rescore started again.", "votes": 5 } ] }, { "id": 1931914, "postDate": "2022-09-09T05:43:42.363Z", "content": "7 out of 7 with internal scoring error
", "rawMarkdown": "7 out of 7 with internal scoring error", "votes": 1 }, { "id": 1932100, "postDate": "2022-09-09T10:10:49.803Z", "content": "Updated: the 3 submissions that were scored successfully yesterday now get errors, and the rest of submissions are scored.
", "rawMarkdown": "Updated: the 3 submissions that were scored successfully yesterday now get errors, and the rest of submissions are scored." }, { "id": 1932183, "postDate": "2022-09-09T12:19:06.843Z", "content": "The rescore is still ongoing. We're about halfway there. The large size of the submission files means it's taking a while to work through. I'll post an update once it's done.
", "rawMarkdown": "The rescore is still ongoing. We're about halfway there. The large size of the submission files means it's taking a while to work through. I'll post an update once it's done.", "votes": 3 } ] }, { "id": 1931702, "postDate": "2022-09-09T00:26:27.903Z", "content": "the new test_cite_targets.h5 file has the following weirdnesses:
\n1) it has 22085 columns (genes), the other two input h5 dataframes have 22050
\n2) the dtypes in the dataframe are a mix of float32 and float64
@ryanholbrook , Can you clarify this?
\nThese columns are in the new file that are not in the other:
\n['ENSG00000255655_AC007570.1',
\n 'ENSG00000213237_AC008155.1',
\n 'ENSG00000273398_AC017083.3',
\n 'ENSG00000233849_AC022201.2',
\n 'ENSG00000231401_AC023481.1',
\n 'ENSG00000263126_AC040162.3',
\n 'ENSG00000285876_AC093912.1',
\n 'ENSG00000275585_AC241377.2',
\n 'ENSG00000272175_AL050343.2',
\n 'ENSG00000232290_AL133260.2',
\n 'ENSG00000267559_AL158154.2',
\n 'ENSG00000273481_AL391069.4',
\n 'ENSG00000161609_CCDC155',
\n 'ENSG00000150394_CDH8',
\n 'ENSG00000233932_CTXN2',
\n 'ENSG00000198077_CYP2A7',
\n 'ENSG00000220871_DNAJC19P6',
\n 'ENSG00000267336_EIF4A2P1',
\n 'ENSG00000251254_GTF2F2P1',
\n 'ENSG00000225978_HAR1A',
\n 'ENSG00000131097_HIGD1B',
\n 'ENSG00000211897_IGHG3',
\n 'ENSG00000233080_LINC01399',
\n 'ENSG00000236849_LINC01474',
\n 'ENSG00000269904_MAP2K4P1',
\n 'ENSG00000266698_MIR3945',
\n 'ENSG00000144227_NXPH2',
\n 'ENSG00000182057_OGFRP1',
\n 'ENSG00000251107_PDCD5P2',
\n 'ENSG00000226964_RHEBP2',
\n 'ENSG00000214988_RPL7AP26',
\n 'ENSG00000248873_SERBP1P6',
\n 'ENSG00000212588_SNORA26',
\n 'ENSG00000166984_TCP10L2',
\n 'ENSG00000226051_ZNF503-AS1']
Thanks for the heads up, @agenlu . I'll investigate.
", "rawMarkdown": "Thanks for the heads up, @agenlu . I'll investigate." }, { "id": 1935340, "postDate": "2022-09-12T03:48:22.737Z", "content": "@splatai where's the test_cite_targets.h5 file? I can't find it in the competition data! Do I miss something here?
", "rawMarkdown": "@splatai where's the test_cite_targets.h5 file? I can't find it in the competition data! Do I miss something here?" }, { "id": 1936527, "postDate": "2022-09-12T20:08:47.723Z", "content": "@minhtu123 I believe this was a typo. The targets for the test data aren't publically available.
\n@agenlu I checked in with the host. You can ignore these columns, if you like.
", "rawMarkdown": "@minhtu123 I believe this was a typo. The targets for the test data aren't publically available.\n\n@agenlu I checked in with the host. You can ignore these columns, if you like." } ] }, { "id": 1932303, "postDate": "2022-09-09T13:48:34.043Z", "content": "Please remember to update it also on Saturn Cloud! Thank you!
", "rawMarkdown": "Please remember to update it also on Saturn Cloud! Thank you!", "votes": 3, "replies": [ { "id": 1962280, "postDate": "2022-09-29T15:53:36.347Z", "content": "I was wondering about this too. @danielburkhardt Have files in Saturn Cloud been updated? The original .h5 files seem more than a month old. Please, check the screenshot attached.
\nPS: \"hace un mes\" means \"a month ago\" in spanish. \"hace 13 dias\" means \"13 days ago\".
\n
I'm not sure I understand your question. To the best of my knowledge, you see here the most recent version of the files. As mentioned in the post above, the original data files were never changed. We decided with Kaggle to just add correct cell IDs for the CITE inputs for the public test as an additional data file. If you look in the original files, the test inputs for Donor 2 CITE are the same as the training data because the test data was accidentally overwritten during the data upload to Kaggle.
", "rawMarkdown": "I'm not sure I understand your question. To the best of my knowledge, you see here the most recent version of the files. As mentioned in the post above, the original data files were never changed. We decided with Kaggle to just add correct cell IDs for the CITE inputs for the public test as an additional data file. If you look in the original files, the test inputs for Donor 2 CITE are the same as the training data because the test data was accidentally overwritten during the data upload to Kaggle.", "votes": 1 }, { "id": 1962367, "postDate": "2022-09-29T16:52:10.367Z", "content": "Thank you Daniel. Sorry, I realize the question might be not clear.
\nI noticed some differences between the original training multiome inputs file (which I happen to keep a copy of) and the latest version. I suspect that files on Saturn Cloud have been kept the same as the original ones. However, I have not deep-dived into these differences. I might publish a notebook soon with details, if I find a pattern in the differences (or no notebook at all, if I find that my previous observations were wrong!).
", "rawMarkdown": "Thank you Daniel. Sorry, I realize the question might be not clear.\n\nI noticed some differences between the original training multiome inputs file (which I happen to keep a copy of) and the latest version. I suspect that files on Saturn Cloud have been kept the same as the original ones. However, I have not deep-dived into these differences. I might publish a notebook soon with details, if I find a pattern in the differences (or no notebook at all, if I find that my previous observations were wrong!)." } ] }, { "id": 1943707, "postDate": "2022-09-17T17:54:27.650Z", "content": "As i understand, train_cite_inputs.h5 still contains test_cite_inputs_day_2_donor_27678.h5 ?
", "rawMarkdown": "As i understand, train_cite_inputs.h5 still contains test_cite_inputs_day_2_donor_27678.h5 ?", "votes": 2 }, { "id": 1930793, "postDate": "2022-09-08T07:44:22.230Z", "content": "Please let us know when the rescoring finishes. It's strange to see some of the submissions are throwing internal scoring error.
", "rawMarkdown": "Please let us know when the rescoring finishes. It's strange to see some of the submissions are throwing internal scoring error.", "votes": 2 }, { "id": 1964523, "postDate": "2022-09-30T19:24:23.137Z", "content": "Hi, wondering if possible to use any public gene database information as prior knowledge for this predication task?
", "rawMarkdown": "Hi, wondering if possible to use any public gene database information as prior knowledge for this predication task?" }, { "id": 1949318, "postDate": "2022-09-21T15:59:46.567Z", "content": "Error: test_cite_inputs have cell ids like
\nad5a949989b2,7,31800,EryP,citeseq, present in cite test data, which should not be
\na3cbc5aa0ec3,7,31800,MkP,citeseq
\n……………………………………………………..
How is this possible?
\nwhile I tried to search for the mentioned error where it supposed to be :
\nday 2 / donor 32606 should be present in the test set but not present, instead above error present ie day7/31800 present in public test_cite.
\nPlease clarify
When I compare the citeSeq train input set and test input set, I can not find the common cell ids? I mean according to the error there are common cells in train and test data but when I check indexes of train and test set I can not find common indexes?
", "rawMarkdown": "When I compare the citeSeq train input set and test input set, I can not find the common cell ids? I mean according to the error there are common cells in train and test data but when I check indexes of train and test set I can not find common indexes?" }, { "id": 1937746, "postDate": "2022-09-13T17:12:23.990Z", "content": "Hi Ryan,
\nIs it possible or any is there plan to upload the raw counts matrix for all the data recently? Thanks a lot!
Hi, do we need to download the newdataset through this website? Thanks.
", "rawMarkdown": "Hi, do we need to download the newdataset through this website? Thanks." }, { "id": 1931681, "postDate": "2022-09-08T23:34:16.377Z", "content": "Hi. All my previous submissions give now internal scoring failure.
", "rawMarkdown": "Hi. All my previous submissions give now internal scoring failure." }, { "id": 1931653, "postDate": "2022-09-08T22:28:18.317Z", "content": "Nearly all of my submissions throwing error now 😄 do I need to resubmit
", "rawMarkdown": "Nearly all of my submissions throwing error now 😄 do I need to resubmit " }, { "id": 1931440, "postDate": "2022-09-08T17:04:55.473Z", "content": "We have all our submissions as errors now… please help check the rescoring process and let us know what changes are needed.
\nThanks.
@drpatrickchan The original rescore attempt failed on our end. We're rerunning it now. We'll make sure that any submission that was successful before will be successful again. You shouldn't need to change anything.
\nApologies for the confusion!
", "rawMarkdown": "@drpatrickchan The original rescore attempt failed on our end. We're rerunning it now. We'll make sure that any submission that was successful before will be successful again. You shouldn't need to change anything. \n\nApologies for the confusion!", "votes": 1 } ] }, { "id": 1930680, "postDate": "2022-09-08T06:00:40.227Z", "content": "Greetings, Ryan!
\nIt is great that Kaggle eliminates leak!
\nDo we have to replace some rows in dfs made from test_cite_inputs.h5 & metadata.csv files with data from test_cite_inputs_day_2_donor_27678.h5 and metadata_cite_day_2_donor_27678.csv?
\nIs it correct?
And what does it mean You should not submit predictions for these new cells?
\nShould we use old version of evaluation_ids.csv and do no changes in submission code?
The current version of evaluation_ids.csv is the same as the old version. You shouldn't need to make any changes to submission code.
Thank you for this update. I have a question.
\nWhy are all my files making errors after rescore? We don't need to do anything according to your explanation.
The same, 14 submissions from 23 were scored as an error (internal score error code).
", "votes": 0, "replies": [] }, { "id": 1930490, "author_name": "YM", "author_url": "", "post_date": "2022-09-07T23:22:58.733000", "content": "The same, 6 submissions from 11were scored as an error .Is there a maximum of 5 re-scoring?
", "votes": 0, "replies": [] }, { "id": 1930492, "author_name": "Sean McManus", "author_url": "", "post_date": "2022-09-07T23:30:52.147000", "content": "Same issue here, I don't care to count how many of ours have errored, needless to say it's the majority of them.
", "votes": 0, "replies": [] }, { "id": 1930494, "author_name": "ONODERA", "author_url": "", "post_date": "2022-09-07T23:54:29.410000", "content": "Almost 100 participants including me were gone. I think it's same issue as mine.
", "votes": 1, "replies": [] }, { "id": 1930687, "author_name": "Jie Wu", "author_url": "", "post_date": "2022-09-08T06:04:58.660000", "content": "Too many submissions get errors. This is non-code competition, imagine what will happen if this is a ode-competition. This makes me wonder if previous code competitions had similar issues as this one or not.
", "votes": 0, "replies": [] }, { "id": 1930760, "author_name": "something4kag", "author_url": "", "post_date": "2022-09-08T07:07:59.313000", "content": "It is possible some will show as Error until the rescoring is complete. Still have some error but some have scores and all are versions of the same notebook. Have seen this before in other competitions that had rescoring happen, so hopefully all will be OK when it is finished.
", "votes": 0, "replies": [] }, { "id": 1930797, "author_name": "minhtu.mt.mt", "author_url": "", "post_date": "2022-09-08T07:50:37.777000", "content": "I think the rescoring is still in progress. Some of my submissions show Error this morning and turn to score later. I also noticed that the top 1 climbed from 6-7th to 1st place without any submission last 10 hours (he was 1st before the rescoring, drop to 6-7th and then return to his place)
", "votes": 0, "replies": [] }, { "id": 1931090, "author_name": "Ryan Holbrook", "author_url": "", "post_date": "2022-09-08T12:59:33.193000", "content": "Apologies for the confusion. It looks like the rescore failed about halfway through. (It may take awhile to work through all the submissions.) I'll run it again and alert our engineering team.
", "votes": 5, "replies": [] }, { "id": 1931112, "author_name": "Zhijian Li", "author_url": "", "post_date": "2022-09-08T13:05:45.227000", "content": "I have only 3 submissions scored sucessfully. All others got errors.
", "votes": 0, "replies": [ { "id": 1931142, "author_name": "Ryan Holbrook", "author_url": "", "post_date": "2022-09-08T13:27:59.283000", "content": "We'll make sure that all submissions that were successful before remain successful. I'm working on getting the rescore started again.
", "votes": 5, "replies": [] } ] }, { "id": 1931914, "author_name": "agenlu", "author_url": "", "post_date": "2022-09-09T05:43:42.363000", "content": "7 out of 7 with internal scoring error
", "votes": 1, "replies": [] }, { "id": 1932100, "author_name": "Zhijian Li", "author_url": "", "post_date": "2022-09-09T10:10:49.803000", "content": "Updated: the 3 submissions that were scored successfully yesterday now get errors, and the rest of submissions are scored.
", "votes": 0, "replies": [] }, { "id": 1932183, "author_name": "Ryan Holbrook", "author_url": "", "post_date": "2022-09-09T12:19:06.843000", "content": "The rescore is still ongoing. We're about halfway there. The large size of the submission files means it's taking a while to work through. I'll post an update once it's done.
", "votes": 3, "replies": [] } ] }, { "id": 1931702, "author_name": "splat.ai", "author_url": "", "post_date": "2022-09-09T00:26:27.903000", "content": "the new test_cite_targets.h5 file has the following weirdnesses:
\n1) it has 22085 columns (genes), the other two input h5 dataframes have 22050
\n2) the dtypes in the dataframe are a mix of float32 and float64
@ryanholbrook , Can you clarify this?
\nThese columns are in the new file that are not in the other:
\n['ENSG00000255655_AC007570.1',
\n 'ENSG00000213237_AC008155.1',
\n 'ENSG00000273398_AC017083.3',
\n 'ENSG00000233849_AC022201.2',
\n 'ENSG00000231401_AC023481.1',
\n 'ENSG00000263126_AC040162.3',
\n 'ENSG00000285876_AC093912.1',
\n 'ENSG00000275585_AC241377.2',
\n 'ENSG00000272175_AL050343.2',
\n 'ENSG00000232290_AL133260.2',
\n 'ENSG00000267559_AL158154.2',
\n 'ENSG00000273481_AL391069.4',
\n 'ENSG00000161609_CCDC155',
\n 'ENSG00000150394_CDH8',
\n 'ENSG00000233932_CTXN2',
\n 'ENSG00000198077_CYP2A7',
\n 'ENSG00000220871_DNAJC19P6',
\n 'ENSG00000267336_EIF4A2P1',
\n 'ENSG00000251254_GTF2F2P1',
\n 'ENSG00000225978_HAR1A',
\n 'ENSG00000131097_HIGD1B',
\n 'ENSG00000211897_IGHG3',
\n 'ENSG00000233080_LINC01399',
\n 'ENSG00000236849_LINC01474',
\n 'ENSG00000269904_MAP2K4P1',
\n 'ENSG00000266698_MIR3945',
\n 'ENSG00000144227_NXPH2',
\n 'ENSG00000182057_OGFRP1',
\n 'ENSG00000251107_PDCD5P2',
\n 'ENSG00000226964_RHEBP2',
\n 'ENSG00000214988_RPL7AP26',
\n 'ENSG00000248873_SERBP1P6',
\n 'ENSG00000212588_SNORA26',
\n 'ENSG00000166984_TCP10L2',
\n 'ENSG00000226051_ZNF503-AS1']
Thanks for the heads up, @agenlu . I'll investigate.
", "votes": 0, "replies": [] }, { "id": 1935340, "author_name": "minhtu.mt.mt", "author_url": "", "post_date": "2022-09-12T03:48:22.737000", "content": "@splatai where's the test_cite_targets.h5 file? I can't find it in the competition data! Do I miss something here?
", "votes": 0, "replies": [] }, { "id": 1936527, "author_name": "Ryan Holbrook", "author_url": "", "post_date": "2022-09-12T20:08:47.723000", "content": "@minhtu123 I believe this was a typo. The targets for the test data aren't publically available.
\n@agenlu I checked in with the host. You can ignore these columns, if you like.
", "votes": 0, "replies": [] } ] }, { "id": 1932303, "author_name": "ioni", "author_url": "", "post_date": "2022-09-09T13:48:34.043000", "content": "Please remember to update it also on Saturn Cloud! Thank you!
", "votes": 3, "replies": [ { "id": 1962280, "author_name": "Alejo Keuro", "author_url": "", "post_date": "2022-09-29T15:53:36.347000", "content": "I was wondering about this too. @danielburkhardt Have files in Saturn Cloud been updated? The original .h5 files seem more than a month old. Please, check the screenshot attached.
\nPS: \"hace un mes\" means \"a month ago\" in spanish. \"hace 13 dias\" means \"13 days ago\".
\n
I'm not sure I understand your question. To the best of my knowledge, you see here the most recent version of the files. As mentioned in the post above, the original data files were never changed. We decided with Kaggle to just add correct cell IDs for the CITE inputs for the public test as an additional data file. If you look in the original files, the test inputs for Donor 2 CITE are the same as the training data because the test data was accidentally overwritten during the data upload to Kaggle.
", "votes": 1, "replies": [] }, { "id": 1962367, "author_name": "Alejo Keuro", "author_url": "", "post_date": "2022-09-29T16:52:10.367000", "content": "Thank you Daniel. Sorry, I realize the question might be not clear.
\nI noticed some differences between the original training multiome inputs file (which I happen to keep a copy of) and the latest version. I suspect that files on Saturn Cloud have been kept the same as the original ones. However, I have not deep-dived into these differences. I might publish a notebook soon with details, if I find a pattern in the differences (or no notebook at all, if I find that my previous observations were wrong!).
", "votes": 0, "replies": [] } ] }, { "id": 1943707, "author_name": "Mirhad", "author_url": "", "post_date": "2022-09-17T17:54:27.650000", "content": "As i understand, train_cite_inputs.h5 still contains test_cite_inputs_day_2_donor_27678.h5 ?
", "votes": 2, "replies": [] }, { "id": 1930793, "author_name": "tarick.morty", "author_url": "", "post_date": "2022-09-08T07:44:22.230000", "content": "Please let us know when the rescoring finishes. It's strange to see some of the submissions are throwing internal scoring error.
", "votes": 2, "replies": [] }, { "id": 1964523, "author_name": "yjyang", "author_url": "", "post_date": "2022-09-30T19:24:23.137000", "content": "Hi, wondering if possible to use any public gene database information as prior knowledge for this predication task?
", "votes": 0, "replies": [] }, { "id": 1949318, "author_name": "its human thing", "author_url": "", "post_date": "2022-09-21T15:59:46.567000", "content": "Error: test_cite_inputs have cell ids like
\nad5a949989b2,7,31800,EryP,citeseq, present in cite test data, which should not be
\na3cbc5aa0ec3,7,31800,MkP,citeseq
\n……………………………………………………..
How is this possible?
\nwhile I tried to search for the mentioned error where it supposed to be :
\nday 2 / donor 32606 should be present in the test set but not present, instead above error present ie day7/31800 present in public test_cite.
\nPlease clarify
When I compare the citeSeq train input set and test input set, I can not find the common cell ids? I mean according to the error there are common cells in train and test data but when I check indexes of train and test set I can not find common indexes?
", "votes": 0, "replies": [] }, { "id": 1937746, "author_name": "Chaozhong Liu", "author_url": "", "post_date": "2022-09-13T17:12:23.990000", "content": "Hi Ryan,
\nIs it possible or any is there plan to upload the raw counts matrix for all the data recently? Thanks a lot!
Hi, do we need to download the newdataset through this website? Thanks.
", "votes": 0, "replies": [] }, { "id": 1931681, "author_name": "Juan Smith Perera", "author_url": "", "post_date": "2022-09-08T23:34:16.377000", "content": "Hi. All my previous submissions give now internal scoring failure.
", "votes": 0, "replies": [] }, { "id": 1931653, "author_name": "Gaurav Rawat", "author_url": "", "post_date": "2022-09-08T22:28:18.317000", "content": "Nearly all of my submissions throwing error now 😄 do I need to resubmit
", "votes": 0, "replies": [] }, { "id": 1931440, "author_name": "Patrick Chan", "author_url": "", "post_date": "2022-09-08T17:04:55.473000", "content": "We have all our submissions as errors now… please help check the rescoring process and let us know what changes are needed.
\nThanks.
@drpatrickchan The original rescore attempt failed on our end. We're rerunning it now. We'll make sure that any submission that was successful before will be successful again. You shouldn't need to change anything.
\nApologies for the confusion!
", "votes": 1, "replies": [] } ] }, { "id": 1930680, "author_name": "Oleg Khudyakov", "author_url": "", "post_date": "2022-09-08T06:00:40.227000", "content": "Greetings, Ryan!
\nIt is great that Kaggle eliminates leak!
\nDo we have to replace some rows in dfs made from test_cite_inputs.h5 & metadata.csv files with data from test_cite_inputs_day_2_donor_27678.h5 and metadata_cite_day_2_donor_27678.csv?
\nIs it correct?
And what does it mean You should not submit predictions for these new cells?
\nShould we use old version of evaluation_ids.csv and do no changes in submission code?
The current version of evaluation_ids.csv is the same as the old version. You shouldn't need to make any changes to submission code.