# DNA

## References

### Final Project
+ 🔗 [**Group 14 Proposal**](https://docs.google.com/document/d/1CGB8yD2HIMu5pKVhnQbqwxZgSlFjz6UVuXEDs7j6V8w/edit?usp=drivesdk)
+ 🔗 [**Group 14 Slide**](https://www.canva.com/design/DAG8PjtrCS0/RfcEWhAMs-PhxHxIbJ51rg/edit?utm_content=DAG8PjtrCS0&utm_campaign=designshare&utm_medium=link2&utm_source=sharebutton)
+ 🔗 [**Group 14 Paper**](https://1drv.ms/w/c/b5bbad118354394f/IQD0jzM7nasiQYrBN7y6e06mAVT0SG7AThqkeodO1MgNWTQ?e=zjBnrj)
+ 🔗 [**IEEE Paper Format**](https://www.scribbr.com/ieee/ieee-paper-format/)
+ 🔗 [**GitHub - Experiments (RogelioKG)**](https://github.com/RogelioKG/test-dna)
+ 🔗 [**GitHub - Experiments (Jay Wu)**](https://github.com/Jay910066/ML_final)

### Paper
+ 🔗 [**Nature Biotechnology - Nucleotide Transformer: building and evaluating robust DNA sequence models**](https://www.nature.com/articles/s41592-024-02523-z)
+ 🔗 [**GitHub - Nucleotide Transformer**](https://github.com/instadeepai/nucleotide-transformer)
+ 🔗 [**GitHub - nucleotide_transformer.md**](https://github.com/instadeepai/nucleotide-transformer/blob/main/docs/nucleotide_transformer.md)
+ 🔗 [**Bioinformatics - DNABERT: pre-trained Bidirectional Encoder Representations from Transformers for DNA sequences**](https://academic.oup.com/bioinformatics/article/37/15/2112/6128680)
+ 🔗 [**GitHub - DNABERT**](https://github.com/jerryji1993/DNABERT)

## Goals
+ 模型：Nucleotide Transformer
+ 資料集：[DNA Sequence Prediction](https://www.kaggle.com/datasets/harshvardhan21/dna-sequence-prediction/code)

## Note
+ 生物小教室 (複習)
1. DNA 序列以 3 個鹼基為一組 (3-mer) 被轉錄成 mRNA
2. mRNA 的每個密碼子 (3-mer) 都會被對應的 tRNA 反密碼子辨識
(欸都，這裡有 wobble base pairing 的問題，但我們先不談)
3. 每種 tRNA 會攜帶一種特定的胺基酸
4. 也就是說某段 DNA 的 3-mer 組成，會影響它最終傾向產生哪類胺基酸
5. 也就間接決定了這段 DNA 序列更可能產生出哪種胺基酸鏈 (蛋白質)
6. 蛋白質，其中有不少是酶，具有生物活性，能藉此推斷出最終可能具備的生物功能
7. 也因此 3-mer 確實是一個推斷出 label 的重要 feature

## Features
| 欄位 | 說明 | 類型 |
| --- | --- | --- |
| `Index` | 索引 | Numerical |
| `NCBIGeneID` | NCBI 基因資料庫 ID | Numerical |
| `Symbol` | 基因名稱 | Text |
| `Description` | 基因描述 | Text |
| `GeneGroupMethod` | 基因分組方法 | Categorical |
| `NucleotideSequence` | **[關鍵]** 核苷酸序列 | Text |

## Labels
| 欄位 | 說明 | 類型 |
| --- | --- | --- |
| `GeneType` | **[關鍵]** 基因功能 | Categorical |

| GeneType | 占比 |
| --- | --- |
| `PSEUDO` | 45.64% |
| `BIOLOGICAL_REGION` | 31.84% |
| `ncRNA` | 10.74% |
| `snoRNA` | 4.86% |
| `PROTEIN_CODING` | 2.21% |
| `tRNA` | 1.78% |
| `OTHER` | 1.60% |
| `rRNA` | 0.86% |
| `snRNA` | 0.46% |
| `scRNA` | 0.01% |