![](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F9711370%2F47c91ebb060a60d80a65cad72bfd79e7%2FAmino%20Group.png?generation=1688372939969645&alt=media\")
![](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F9711370%2F8b56b794e2665247b9e3d9ef55844301%2FCarboxyl%20Group.png?generation=1688373003090699&alt=media\")
Hey guys, the goal of this discussion is to provide efficient resources to new-comers so that they can catch up with the competition. As I am also exploring the competition, this discussion will be updated as better resources are available.
This discussion is divided into distinct sections, to improve feasibility
Multilabel-Classification problemIncreasing the number of Neurons/Network Layers, will not provide a significant impact on the results, but rather cost a lot more with GPUTrained and Evaluated, TEMPROT, took less training time and gave better results.The competition is about predicting the function of Proteins in the Body
Proteins are made up of long chains of Amino-Acids connected with the help of Peptide-Bonds. Further Amino Acids are made up of mainly 2 molecules named as
| \n | \n |
|---|---|
| Amino Group | \n | \n
| Carboxyl Group | \n | \n
Amino Acid also have a Side Chain, which introduces the possibility of different types of Amino Acids. Different types of Amino Acids group/connect together to form different types of Proteins. Different types of Proteins conduct different Biological Function in the body.
Our goal is to predict these Biological Functions.
\nGene Ontology is a controlled vocabulary that describes the functions of genes and gene products.
There are mainly 3 Ontologies
| \n | \n | \n |
|---|---|---|
| Biological Process | \nDescribes the biological processes | \nA gene product might be involved in the process of cell cycle/signal transduction. | \n
| Cellular Component | \nDescribes the cellular components | \nA gene product might be located in the nucleus/cytoplasm. | \n
| Molecular Function | \nDescribes the molecular functions | \nA gene product might be involved in the catalysis of a reaction/binding of a molecule. | \n
We are given the GO-Terms and we need to predict the Biological Functions.
\nGo-Basic.obo file has the information of GO-terms particular with Ontologies
subontology_roots = {'BPO':'GO:0',\n 'CCO':'GO:5',\n 'MFO':'GO:4'}\nWhen read/printined
[Term]\n GO:\n mitochondrion inheritance\n biological_process\n [GOC:mcc, PMID:, PMID:]\n EXACT []\n GO: ! organelle inheritance\n GO: ! mitochondrion distribution\nTrain-Fasta.fasta is a Fasta Sequence conatining the Protein Sequences which can be read
Bio SeqIO\n\nfasta_seq = SeqIO.parse((\"/kaggle/input/cafa-5-protein-function-prediction/Train/train_sequences.fasta\") ,)\n
\nUpon getting deep, the Sequence looks like this
(SeqRecord(=Seq('HCISSLKLTAFFKRSFLLSPEKHLVLLRDGRTLIGFLRSIDQFANLVLHQTVER...DEY'), =, =, =, dbxrefs=[])\nThese can be distincitlvely reached by functions such as sequence.seq/sequence.id, where sequence is a object of the Fasta Files. We can get the object by iterating over the Fasta Files as
for fasta_seq:\nTrain-Terms.tsv is a tsv file containing the main training datalike this
| \n | EntryID | \nterm | \naspect | \n
|---|---|---|---|
| 0 | \nA0A009IHW8 | \nGO:0008152 | \nBPO | \n
| 1 | \nA0A009IHW8 | \nGO:0034655 | \nBPO | \n
| 2 | \nA0A009IHW8 | \nGO:0072523 | \nBPO | \n
| 3 | \nA0A009IHW8 | \nGO:0044270 | \nBPO | \n
| 4 | \nA0A009IHW8 | \nGO:0006753 | \nBPO | \n
| … | \n… | \n… | \n… | \n
The Test Set is unknown at the beginning of the competition. It will contain protein sequences their functions from the test superset that gained experimental annotations between the submission-deadline and the time of evaluation.
Here is the Pytorch Dataloader with O(1) Time-Complexity and O(n) Space-Complexity. This uses the following additional datasets
| \n | \n |
|---|---|
| train_targets_top500 | \nTop 500 targets | \n
numpy np\n\n torch\n torch.utils.data Dataset\n\n ():\n\n ():\n (Pytorch_Dataset).__init__()\n\n self.datatype = datatype\n self.embeds = np.load()\n\n datatype == :\n\n self.targets = np.load()[:self.embeds.shape[]]\n\n (): self.targets.shape[]\n\n ():\n\n r_embed = torch.tensor(self.embeds[index] , dtype = torch.float32)\n\n self.datatype == :\n\n r_targets = torch.tensor(self.targets[index], dtype = torch.float32)\n\n r_embed, r_targets\n\n r_embed\nProteiNet 🧬 PyTorch+EMS2/T5/ProtBERT Embedding - Henri Upton
TEMPROT: protein function annotation using transformers embeddings and homology search
Hey guys, the goal of this discussion is to provide efficient resources to new-comers so that they can catch up with the competition. As I am also exploring the competition, this discussion will be updated as better resources are available.
This discussion is divided into distinct sections, to improve feasibility
Multilabel-Classification problemIncreasing the number of Neurons/Network Layers, will not provide a significant impact on the results, but rather cost a lot more with GPUTrained and Evaluated, TEMPROT, took less training time and gave better results.The competition is about predicting the function of Proteins in the Body
Proteins are made up of long chains of Amino-Acids connected with the help of Peptide-Bonds. Further Amino Acids are made up of mainly 2 molecules named as
| \n | \n |
|---|---|
| Amino Group | \n | \n
| Carboxyl Group | \n | \n
Amino Acid also have a Side Chain, which introduces the possibility of different types of Amino Acids. Different types of Amino Acids group/connect together to form different types of Proteins. Different types of Proteins conduct different Biological Function in the body.
Our goal is to predict these Biological Functions.
\nGene Ontology is a controlled vocabulary that describes the functions of genes and gene products.
There are mainly 3 Ontologies
| \n | \n | \n |
|---|---|---|
| Biological Process | \nDescribes the biological processes | \nA gene product might be involved in the process of cell cycle/signal transduction. | \n
| Cellular Component | \nDescribes the cellular components | \nA gene product might be located in the nucleus/cytoplasm. | \n
| Molecular Function | \nDescribes the molecular functions | \nA gene product might be involved in the catalysis of a reaction/binding of a molecule. | \n
We are given the GO-Terms and we need to predict the Biological Functions.
\nGo-Basic.obo file has the information of GO-terms particular with Ontologies
subontology_roots = {'BPO':'GO:0',\n 'CCO':'GO:5',\n 'MFO':'GO:4'}\nWhen read/printined
[Term]\n GO:\n mitochondrion inheritance\n biological_process\n [GOC:mcc, PMID:, PMID:]\n EXACT []\n GO: ! organelle inheritance\n GO: ! mitochondrion distribution\nTrain-Fasta.fasta is a Fasta Sequence conatining the Protein Sequences which can be read
Bio SeqIO\n\nfasta_seq = SeqIO.parse((\"/kaggle/input/cafa-5-protein-function-prediction/Train/train_sequences.fasta\") ,)\n
\nUpon getting deep, the Sequence looks like this
(SeqRecord(=Seq('HCISSLKLTAFFKRSFLLSPEKHLVLLRDGRTLIGFLRSIDQFANLVLHQTVER...DEY'), =, =, =, dbxrefs=[])\nThese can be distincitlvely reached by functions such as sequence.seq/sequence.id, where sequence is a object of the Fasta Files. We can get the object by iterating over the Fasta Files as
for fasta_seq:\nTrain-Terms.tsv is a tsv file containing the main training datalike this
| \n | EntryID | \nterm | \naspect | \n
|---|---|---|---|
| 0 | \nA0A009IHW8 | \nGO:0008152 | \nBPO | \n
| 1 | \nA0A009IHW8 | \nGO:0034655 | \nBPO | \n
| 2 | \nA0A009IHW8 | \nGO:0072523 | \nBPO | \n
| 3 | \nA0A009IHW8 | \nGO:0044270 | \nBPO | \n
| 4 | \nA0A009IHW8 | \nGO:0006753 | \nBPO | \n
| … | \n… | \n… | \n… | \n
The Test Set is unknown at the beginning of the competition. It will contain protein sequences their functions from the test superset that gained experimental annotations between the submission-deadline and the time of evaluation.
Here is the Pytorch Dataloader with O(1) Time-Complexity and O(n) Space-Complexity. This uses the following additional datasets
| \n | \n |
|---|---|
| train_targets_top500 | \nTop 500 targets | \n
numpy np\n\n torch\n torch.utils.data Dataset\n\n ():\n\n ():\n (Pytorch_Dataset).__init__()\n\n self.datatype = datatype\n self.embeds = np.load()\n\n datatype == :\n\n self.targets = np.load()[:self.embeds.shape[]]\n\n (): self.targets.shape[]\n\n ():\n\n r_embed = torch.tensor(self.embeds[index] , dtype = torch.float32)\n\n self.datatype == :\n\n r_targets = torch.tensor(self.targets[index], dtype = torch.float32)\n\n r_embed, r_targets\n\n r_embed\nProteiNet 🧬 PyTorch+EMS2/T5/ProtBERT Embedding - Henri Upton
TEMPROT: protein function annotation using transformers embeddings and homology search
\n|Carboxyl Group|\n\nAmino Acid also have a `Side Chain`, which introduces the possibility of `different types of Amino Acids`. `Different types of Amino Acids` group/connect together to form `different types of Proteins`. `Different types of Proteins` conduct `different Biological Function` in the body. \n\nOur goal is to predict these Biological Functions.\n\n****\n\nGene Ontology is a `controlled vocabulary` that describes the `functions of genes` and gene products. \n\nThere are mainly 3 `Ontologies`\n\n||||\n|---|---|---\n|Biological Process|Describes the `biological processes`|A gene product might be involved in the process of `cell cycle`/`signal transduction.`\n|Cellular Component|Describes the `cellular components`|A gene product might be located in the `nucleus`/`cytoplasm`.\n|Molecular Function|Describes the `molecular functions`|A gene product might be involved in the `catalysis of a reaction`/`binding of a molecule`.\n\nWe are given the GO-Terms and we need to predict the Biological Functions. \n\n# 3 | Data\n****\n`Go-Basic.obo` file has the information of `GO-terms particular with Ontologies` \n```\nsubontology_roots = {'BPO':'GO:0008150',\n 'CCO':'GO:0005575',\n 'MFO':'GO:0003674'}\n```\nWhen `read/printined`\n```\n[Term]\nid: GO:0000001\nname: mitochondrion inheritance\nnamespace: biological_process\ndef: \"The distribution of mitochondria, including the mitochondrial genome, into daughter cells after mitosis or meiosis, mediated by interactions between mitochondria and the cytoskeleton.\" [GOC:mcc, PMID:10873824, PMID:11389764]\nsynonym: \"mitochondrial inheritance\" EXACT []\nis_a: GO:0048308 ! organelle inheritance\nis_a: GO:0048311 ! mitochondrion distribution\n```\n****\n`Train-Fasta.fasta` is a `Fasta` Sequence conatining the Protein Sequences which can be read\n```\nfrom Bio import SeqIO\n\nfasta_seq = SeqIO.parse(open(\"/kaggle/input/cafa-5-protein-function-prediction/Train/train_sequences.fasta\") ,'fasta')\n``` \nUpon getting deep, the Sequence looks like this \n```\n(SeqRecord(seq=Seq('HCISSLKLTAFFKRSFLLSPEKHLVLLRDGRTLIGFLRSIDQFANLVLHQTVER...DEY'), id='A0A8I6GHU0', name='A0A8I6GHU0', description='A0A8I6GHU0 tr|A0A8I6GHU0|A0A8I6GHU0_RAT U6 snRNA-associated Sm-like protein LSm1 OS=Rattus norvegicus OX=10116 GN=Lsm1 PE=3 SV=1', dbxrefs=[])\n```\nThese can be distincitlvely reached by functions such as `sequence.seq`/`sequence.id`, where sequence is a object of the Fasta Files. We can get the object by iterating over the Fasta Files as \n```\nfor sequence in fasta_seq:pass\n```\n****\n`Train-Terms.tsv` is a tsv file containing the `main training data `like this \n\n||EntryID|term|aspect\n|---|---|---|---\n|0|A0A009IHW8|GO:0008152|BPO\n|1|A0A009IHW8|GO:0034655|BPO\n|2|A0A009IHW8|GO:0072523|BPO\n|3|A0A009IHW8|GO:0044270|BPO\n|4|A0A009IHW8|GO:0006753|BPO\n|...|...|...|...\n****\nThe `Test` `Set` is `unknown at the beginning` of the competition. It will contain `protein sequences` `their functions` from the `test superset` that `gained experimental annotations` between the `submission-deadline` and the `time of evaluation`.\n\n# 4 | Pytorch DataLoader \nHere is the `Pytorch Dataloader` with `O(1)` `Time-Complexity` and `O(n)` `Space-Complexity`. This uses the following additional datasets\n\n|||\n|---|---\n|[train_targets_top500](https://www.kaggle.com/datasets/siddhvr/train-targets-top500)|Top 500 targets\n\n```\nimport numpy as np\n\nimport torch\nfrom torch.utils.data import Dataset\n\nclass Pytorch_Dataset(Dataset):\n\n def __init__(self , datatype ):\n super(Pytorch_Dataset).__init__()\n\n self.datatype = datatype\n self.embeds = np.load(\"/kaggle/input/cafa-sample-embeddings/Embeddings/T5-Prot-BERT BFD/Train/Embeds.npy\")\n \n if datatype == \"train\":\n\n self.targets = np.load(\"/kaggle/input/train-targets-top500/train_targets_top500.npy\")[:self.embeds.shape[0]]\n \n def __len__(self):return self.targets.shape[0]\n\n def __getitem__(self , index):\n\n r_embed = torch.tensor(self.embeds[index] , dtype = torch.float32)\n\n if self.datatype == \"train\":\n\n r_targets = torch.tensor(self.targets[index], dtype = torch.float32)\n\n return r_embed, r_targets\n\n return r_embed\n```\n\n# 5 | Sources \n\n### Data\n\n* [CAFA 5 protein function with TensorFlow](https://www.kaggle.com/code/gusthema/cafa-5-protein-function-with-tensorflow) - [Gusthema](https://www.kaggle.com/gusthema),[elliot robot](https://www.kaggle.com/elliot2robot)\n\n### Pytorch DataLoader \n\n* [ProteiNet 🧬 PyTorch+EMS2/T5/ProtBERT Embedding](https://www.kaggle.com/code/henriupton/proteinet-pytorch-ems2-t5-protbert-embeddings) - [Henri Upton](https://www.kaggle.com/henriupton)\n\n* [TEMPROT: protein function annotation using transformers embeddings and homology search](https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-023-05375-0)",
"votes": null
},
{
"id": "2328483",
"postDate": "07/03/2023 15:59:25",
"content": "@ayushs9020 hi
\nDo you like to come to zoom meeting now:):
https://us02web.zoom.us/j/84239724288?pwd=L3hQSXRTaUppWlpmWEs5SUFzMitUZz09
\n?
", "rawMarkdown": "ayushs9020 hi\nDo you like to come to zoom meeting now:):\n\nhttps://us02web.zoom.us/j/84239724288?pwd=L3hQSXRTaUppWlpmWEs5SUFzMitUZz09\n\n?", "votes": null }, { "id": "2329250", "postDate": "07/04/2023 06:59:19", "content": "Hi @liudacheldieva , sorry for replying you late, Kaggle didn't notified me.
\nActually I am currently not in a situation to attend a zoom meeting, because of several reasons. It would be great if we could communicate over Mail/Linkedin.
Hope you will not mind :)
", "rawMarkdown": "Hi @liudacheldieva , sorry for replying you late, Kaggle didn't notified me.\n\nActually I am `currently not in a situation to attend a zoom meeting`, because of several reasons. It would be great if `we could communicate` over `Mail`/`Linkedin`.\n\n* Mail - ayushsinghal659@gmail.com\n* Linkedin - [Ayush Singhal](https://www.linkedin.com/in/ayush-singhal-a0614a221/)\n\nHope you will not mind :)", "votes": null }, { "id": "2331596", "postDate": "07/05/2023 16:39:35", "content": "You are welcome to do as you wish, but here is a simple reminder: private communications is generally not allowed except between group members, or at least potential group members. There is a reason why they are organizing public Zoom meetings.
\nYou may want to read the rules section.
\nhttps://www.kaggle.com/competitions/cafa-5-protein-function-prediction/rules
", "rawMarkdown": "You are welcome to do as you wish, but here is a simple reminder: private communications is generally not allowed except between group members, or at least potential group members. There is a reason why they are organizing public Zoom meetings.\n\nYou may want to read the rules section.\n\nhttps://www.kaggle.com/competitions/cafa-5-protein-function-prediction/rules", "votes": null } ], "comments": [ { "id": 2328483, "author_name": "", "author_url": "", "post_date": "07/03/2023 15:59:25", "content": "@ayushs9020 hi
\nDo you like to come to zoom meeting now:):
https://us02web.zoom.us/j/84239724288?pwd=L3hQSXRTaUppWlpmWEs5SUFzMitUZz09
\n?
", "votes": null, "replies": [ { "id": 2329250, "author_name": "ayushs9020", "author_url": "", "post_date": "07/04/2023 06:59:19", "content": "Hi @liudacheldieva , sorry for replying you late, Kaggle didn't notified me.
\nActually I am currently not in a situation to attend a zoom meeting, because of several reasons. It would be great if we could communicate over Mail/Linkedin.
Hope you will not mind :)
", "votes": null, "replies": [ { "id": 2331596, "author_name": "tilii7", "author_url": "", "post_date": "07/05/2023 16:39:35", "content": "You are welcome to do as you wish, but here is a simple reminder: private communications is generally not allowed except between group members, or at least potential group members. There is a reason why they are organizing public Zoom meetings.
\nYou may want to read the rules section.
\nhttps://www.kaggle.com/competitions/cafa-5-protein-function-prediction/rules
", "votes": null, "replies": [] } ] } ] } ], "raw_markdown_by_id": { "2327969": "Hey guys, the goal of this discussion is to `provide efficient resources` to new-comers so that they can catch up with the competition. As I am also exploring the competition, this `discussion will be updated as better resources are available`.\n\nThis discussion is `divided into distinct sections`, to improve feasibility\n\n# 1 | Advisory \n\n* This is a `Multilabel-Classification` problem\n* `Increasing` the number of `Neurons`/`Network Layers`, will `not` provide a `significant impact` on the `results`, but rather `cost` a lot more with `GPU`\n* When `Trained and Evaluated`, `TEMPROT`, took `less training time` and gave `better results.`\n\n# 2 | Field Knowledge\n****\nThe competition is about `predicting the function of Proteins` in the Body\n****\n`Proteins` are made up of `long chains of Amino-Acids connected` with the help of `Peptide-Bonds`. Further Amino Acids are made up of mainly 2 molecules named as \n\n|||\n|---|---\n|Amino Group|\n|Carboxyl Group|\n\nAmino Acid also have a `Side Chain`, which introduces the possibility of `different types of Amino Acids`. `Different types of Amino Acids` group/connect together to form `different types of Proteins`. `Different types of Proteins` conduct `different Biological Function` in the body. \n\nOur goal is to predict these Biological Functions.\n\n****\n\nGene Ontology is a `controlled vocabulary` that describes the `functions of genes` and gene products. \n\nThere are mainly 3 `Ontologies`\n\n||||\n|---|---|---\n|Biological Process|Describes the `biological processes`|A gene product might be involved in the process of `cell cycle`/`signal transduction.`\n|Cellular Component|Describes the `cellular components`|A gene product might be located in the `nucleus`/`cytoplasm`.\n|Molecular Function|Describes the `molecular functions`|A gene product might be involved in the `catalysis of a reaction`/`binding of a molecule`.\n\nWe are given the GO-Terms and we need to predict the Biological Functions. \n\n# 3 | Data\n****\n`Go-Basic.obo` file has the information of `GO-terms particular with Ontologies` \n```\nsubontology_roots = {'BPO':'GO:0008150',\n 'CCO':'GO:0005575',\n 'MFO':'GO:0003674'}\n```\nWhen `read/printined`\n```\n[Term]\nid: GO:0000001\nname: mitochondrion inheritance\nnamespace: biological_process\ndef: \"The distribution of mitochondria, including the mitochondrial genome, into daughter cells after mitosis or meiosis, mediated by interactions between mitochondria and the cytoskeleton.\" [GOC:mcc, PMID:10873824, PMID:11389764]\nsynonym: \"mitochondrial inheritance\" EXACT []\nis_a: GO:0048308 ! organelle inheritance\nis_a: GO:0048311 ! mitochondrion distribution\n```\n****\n`Train-Fasta.fasta` is a `Fasta` Sequence conatining the Protein Sequences which can be read\n```\nfrom Bio import SeqIO\n\nfasta_seq = SeqIO.parse(open(\"/kaggle/input/cafa-5-protein-function-prediction/Train/train_sequences.fasta\") ,'fasta')\n``` \nUpon getting deep, the Sequence looks like this \n```\n(SeqRecord(seq=Seq('HCISSLKLTAFFKRSFLLSPEKHLVLLRDGRTLIGFLRSIDQFANLVLHQTVER...DEY'), id='A0A8I6GHU0', name='A0A8I6GHU0', description='A0A8I6GHU0 tr|A0A8I6GHU0|A0A8I6GHU0_RAT U6 snRNA-associated Sm-like protein LSm1 OS=Rattus norvegicus OX=10116 GN=Lsm1 PE=3 SV=1', dbxrefs=[])\n```\nThese can be distincitlvely reached by functions such as `sequence.seq`/`sequence.id`, where sequence is a object of the Fasta Files. We can get the object by iterating over the Fasta Files as \n```\nfor sequence in fasta_seq:pass\n```\n****\n`Train-Terms.tsv` is a tsv file containing the `main training data `like this \n\n||EntryID|term|aspect\n|---|---|---|---\n|0|A0A009IHW8|GO:0008152|BPO\n|1|A0A009IHW8|GO:0034655|BPO\n|2|A0A009IHW8|GO:0072523|BPO\n|3|A0A009IHW8|GO:0044270|BPO\n|4|A0A009IHW8|GO:0006753|BPO\n|...|...|...|...\n****\nThe `Test` `Set` is `unknown at the beginning` of the competition. It will contain `protein sequences` `their functions` from the `test superset` that `gained experimental annotations` between the `submission-deadline` and the `time of evaluation`.\n\n# 4 | Pytorch DataLoader \nHere is the `Pytorch Dataloader` with `O(1)` `Time-Complexity` and `O(n)` `Space-Complexity`. This uses the following additional datasets\n\n|||\n|---|---\n|[train_targets_top500](https://www.kaggle.com/datasets/siddhvr/train-targets-top500)|Top 500 targets\n\n```\nimport numpy as np\n\nimport torch\nfrom torch.utils.data import Dataset\n\nclass Pytorch_Dataset(Dataset):\n\n def __init__(self , datatype ):\n super(Pytorch_Dataset).__init__()\n\n self.datatype = datatype\n self.embeds = np.load(\"/kaggle/input/cafa-sample-embeddings/Embeddings/T5-Prot-BERT BFD/Train/Embeds.npy\")\n \n if datatype == \"train\":\n\n self.targets = np.load(\"/kaggle/input/train-targets-top500/train_targets_top500.npy\")[:self.embeds.shape[0]]\n \n def __len__(self):return self.targets.shape[0]\n\n def __getitem__(self , index):\n\n r_embed = torch.tensor(self.embeds[index] , dtype = torch.float32)\n\n if self.datatype == \"train\":\n\n r_targets = torch.tensor(self.targets[index], dtype = torch.float32)\n\n return r_embed, r_targets\n\n return r_embed\n```\n\n# 5 | Sources \n\n### Data\n\n* [CAFA 5 protein function with TensorFlow](https://www.kaggle.com/code/gusthema/cafa-5-protein-function-with-tensorflow) - [Gusthema](https://www.kaggle.com/gusthema),[elliot robot](https://www.kaggle.com/elliot2robot)\n\n### Pytorch DataLoader \n\n* [ProteiNet 🧬 PyTorch+EMS2/T5/ProtBERT Embedding](https://www.kaggle.com/code/henriupton/proteinet-pytorch-ems2-t5-protbert-embeddings) - [Henri Upton](https://www.kaggle.com/henriupton)\n\n* [TEMPROT: protein function annotation using transformers embeddings and homology search](https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-023-05375-0)",
"2328483": "ayushs9020 hi\nDo you like to come to zoom meeting now:):\n\nhttps://us02web.zoom.us/j/84239724288?pwd=L3hQSXRTaUppWlpmWEs5SUFzMitUZz09\n\n?",
"2329250": "Hi @liudacheldieva , sorry for replying you late, Kaggle didn't notified me.\n\nActually I am `currently not in a situation to attend a zoom meeting`, because of several reasons. It would be great if `we could communicate` over `Mail`/`Linkedin`.\n\n* Mail - ayushsinghal659@gmail.com\n* Linkedin - [Ayush Singhal](https://www.linkedin.com/in/ayush-singhal-a0614a221/)\n\nHope you will not mind :)",
"2331596": "You are welcome to do as you wish, but here is a simple reminder: private communications is generally not allowed except between group members, or at least potential group members. There is a reason why they are organizing public Zoom meetings.\n\nYou may want to read the rules section.\n\nhttps://www.kaggle.com/competitions/cafa-5-protein-function-prediction/rules"
},
"source": "meta"
}