RNA碱基不成对概率预测竞赛基线模型基于飞桨2.0开发，可预测RNA序列各点位不成对概率，该概率在mRNA疫苗设计等领域有重要应用。提供5000条训练数据，基线用简单网络模型，训练脚本和测试脚本可分别训练模型和预测，开发者可从数据预处理等多方面优化模型。
RNA碱基不成对概率预测的基线模型
RNA结构预测竞赛：RNA碱基不成对概率预测基线模型现已开放，采用非常简单的网络模型，就能得到还不错的结果。欢迎开发者贡献更好的基线作品~
赛题介绍
“RNA碱基不成对概率”衡量了RNA序列在各个点位是否能形成稳定的碱基对（base pair），是RNA结构的重要属性，并可被应用在mRNA疫苗序列设计、药物研发等领域。例如mRNA疫苗序列通常不稳定，而RNA碱基不成对概率较高的点位正是易被降解的位置；又如RNA 碱基不成对概率较高的点位通常更容易与其他RNA序列相互作用，形成RNA-RNA binding等，这一特性也被广泛应用于疾病诊断和RNA药物研发。
本次比赛提供了5000条训练数据，请选手基于训练数据和飞桨平台，开发模型预测RNA碱基不成对概率。
（Tips：机器学习框架方面只允许使用飞桨深度学习框架哦）
竞赛数据集
In [ ]# 检查数据集所在路径!tree /home/aistudio/data登录后复制
基线系统代码结构
本次基线基于飞桨PaddlePaddle2.0版本。
In [3]# 检查源代码文件结构# !cd work; mkdir model!tree /home/aistudio/work -L 2登录后复制
/home/aistudio/work├── data│ ├── dev.txt│ ├── test_nolabel.txt│ └── train.txt├── model│ └── placeholder.txt├── model-0│ └── model_dev=0.0772├── README.txt├── src│ ├── const.py│ ├── dataset.py│ ├── __init__.py│ ├── main.py│ ├── network.py│ ├── utils.py│ └── vocabulary.py├── test_log.txt└── train_log.txt5 directories, 14 files登录后复制登录后复制
训练脚本
python src/main.py train --model-path-base [model_directory_name]
本代码会训练一个模型，并且保存到指定位置，训练日志默认保存到文件train_log.txt
注意：由于初始化的不稳定，可能需要多次训练，比较合理的验证集(dev)均方误差损失值(MSE loss)为0.05-0.08
样例
python src/main.py train --model-path-base model
你将会看到类似如下的训练日志
epoch 1 batch 40 processed 640 batch-loss 0.1984 epoch-elapsed 0h00m10s total-elapsed 0h00m11s epoch 1 batch 41 processed 656 batch-loss 0.2119 epoch-elapsed 0h00m10s total-elapsed 0h00m11s epoch 1 batch 42 processed 672 batch-loss 0.2205 epoch-elapsed 0h00m11s total-elapsed 0h00m11s epoch 1 batch 43 processed 688 batch-loss 0.2128 epoch-elapsed 0h00m11s total-elapsed 0h00m11s # Dev Average Loss: 0.212 (MSE) -> 0.461 (RMSD)登录后复制登录后复制
注意事项
请使用GPU版本的配置环境运行本模块
In [ ]# To train:# python src/main.py train --model-path-base [model_directory_name]!cd work; python src/main.py train --model-path-base model登录后复制登录后复制
预测脚本
python src/main.py test --model-path-base [saved_model_directory]
本代码会预测一个模型，日志和结果默认保存到文件test_log.txt
样例
用不带标签的测试集来预测：python src/main.py test --model-path-base model-0/model_dev\=0.0772/用带标签的测试集来预测并评估:
python src/main.py test_withlabel --model-path-base model-0/model_dev\=0.0772/
样例输出
# python3 src/main.py test_withlabel --model-path-base model-0/model_dev=0.0772Loading data...Loading model...initializing vacabularies... done.Sequence(6): ['登录后复制登录后复制 由于比赛的公开数据不提供测试集的标签，故本基线模型无法运行预设的test_withlabel，除非用户自己生成一个带标签的测试集~/data/test.txt。', ' ', 'A', 'C', 'G', 'U']Brackets(5): [' ', ' ', '(', ')', '.']W0113 21:57:44.871776 221 device_context.cc:252] Please NOTE: device: 0, CUDA Capability: 70, Driver API Version: 11.0, Runtime API Version: 9.0W0113 21:57:44.878015 221 device_context.cc:260] device: 0, cuDNN Version: 7.6.# Dev Average Loss: 0.0772 (MSE) -> 0.2778 (RMSD)# Test Average Loss: 0.0445 (MSE) -> 0.2111 (RMSD)
注意事项
请使用GPU版本的配置环境运行本模块
In [ ]# To test 1:# python src/main.py test --model-path-base [saved_model_directory]!cd work; python src/main.py test --model-path-base model-0/model_dev\=0.0772登录后复制登录后复制 In [ ]
# To test 2:# python src/main.py test_withlabel --model-path-base [saved_model_directory]# 由于比赛的公开数据不提供测试集的标签，故本基线模型无法运行预设的test_withlabel，除非用户自己生成一个带标签的测试集~/data/test.txt#### !cd work; python src/main.py test_withlabel --model-path-base model-0/model_dev\=0.0772登录后复制
写在最后
选手可在基线模型基础上持续改进，也可以采用全新的模型。可以尝试从以下几方面来进行调优:
输入数据预处理，提取更多feature基线模型使用LinearFold预测的RNA二级结构作为辅助feature。选手可以尝试增加更多的辅助feature，如：使用其他二级结构预测软件（如Vienna RNAfold, RNAstructure, CONTRAfold等）生成新的二级结构feature。更复杂的Embedding形式可以尝试在Embedding层使用Elmo, Bert等预训练模型优化网络结构和参数隐层大小选择 - 宽度和层数尝试复杂网络构建尝试正则化、dropout等方式避免过拟合选择学习率等超参数选择合适的损失函数尝试不同的优化器点击跳转到 中文 说明
Baseline for RNA base unpaired probability
RNA structure prediction contest: The baseline model of RNA base unpairing probability prediction is now released. Using a very simple network model, you can get good results. Developers are welcome to contribute better baseline works~
About the challange
"RNA base unpaired probability" measures whether the RNA sequence can form a stable base pair at certain point. It is an important attribute of RNA structure and can be used in the design of mRNA vaccine sequence, drug development and other fields. For example, mRNA vaccine sequences are usually unstable, and the points with a higher probability of unpaired RNA bases are the positions that are easily degraded; another example is the points with higher probability of unpaired RNA bases are usually more likely to interact with other RNA sequences, with the formation of RNA-RNA binding, etc. This feature is also widely used in disease diagnosis and RNA drug development.
This competition provides 5000 training data. You are asked to develop a model to predict the unpaired probability of RNA bases, given the training data and PaddlePaddle platform.
（Tips：Machine learning models can be only developed with PaddlePaddle deep learning frameworks）
Dataset for RNA contest
In [ ]# check dataset!tree /home/aistudio/data登录后复制
Baseline system
This baseline is developed with PaddlePaddle2.0.
In [4]# check codebase# !cd work; mkdir model!tree /home/aistudio/work -L 2登录后复制
/home/aistudio/work├── data│ ├── dev.txt│ ├── test_nolabel.txt│ └── train.txt├── model│ └── placeholder.txt├── model-0│ └── model_dev=0.0772├── README.txt├── src│ ├── const.py│ ├── dataset.py│ ├── __init__.py│ ├── main.py│ ├── network.py│ ├── utils.py│ └── vocabulary.py├── test_log.txt└── train_log.txt5 directories, 14 files登录后复制登录后复制
To train
python src/main.py train --model-path-base [model_directory_name]
This will train a paddle model and save it in the specified directory. Training log will be outputed to train_log.txt by default.
Note: You may need to train a few times, as the model might have bad initialization and may not train well. (A dev MSE loss of about 0.05-0.08 is good)
Example
python src/main.py train --model-path-base model
training log like this
epoch 1 batch 40 processed 640 batch-loss 0.1984 epoch-elapsed 0h00m10s total-elapsed 0h00m11s epoch 1 batch 41 processed 656 batch-loss 0.2119 epoch-elapsed 0h00m10s total-elapsed 0h00m11s epoch 1 batch 42 processed 672 batch-loss 0.2205 epoch-elapsed 0h00m11s total-elapsed 0h00m11s epoch 1 batch 43 processed 688 batch-loss 0.2128 epoch-elapsed 0h00m11s total-elapsed 0h00m11s # Dev Average Loss: 0.212 (MSE) -> 0.461 (RMSD)登录后复制登录后复制
Note
Please use GPU/CUDA configuration to run this block
In [ ]# To train:# python src/main.py train --model-path-base [model_directory_name]!cd work; python src/main.py train --model-path-base model登录后复制登录后复制
To test
python src/main.py test --model-path-base [saved_model_directory]
This will evaluate a trained model. Testing log will be outputed to test_log.txt by default.
Example
if you have blind testset to predict:python src/main.py test --model-path-base model-0/model_dev\=0.0772/if you have full testset (with label) to predict and validate:
python src/main.py test_withlabel --model-path-base model-0/model_dev\=0.0772/
Outpit like this:
# python3 src/main.py test_withlabel --model-path-base model-0/model_dev=0.0772Loading data...Loading model...initializing vacabularies... done.Sequence(6): ['登录后复制登录后复制 Since we don't offer labelled testset in baseline, you may not run test_withlabel unless you create ~/data/test.txt by yourself.', ' ', 'A', 'C', 'G', 'U']Brackets(5): [' ', ' ', '(', ')', '.']W0113 21:57:44.871776 221 device_context.cc:252] Please NOTE: device: 0, CUDA Capability: 70, Driver API Version: 11.0, Runtime API Version: 9.0W0113 21:57:44.878015 221 device_context.cc:260] device: 0, cuDNN Version: 7.6.# Dev Average Loss: 0.0772 (MSE) -> 0.2778 (RMSD)# Test Average Loss: 0.0445 (MSE) -> 0.2111 (RMSD)
Note
Please use GPU/CUDA configuration to run this block
In [ ]# To test 1:# python src/main.py test --model-path-base [saved_model_directory]!cd work; python src/main.py test --model-path-base model-0/model_dev\=0.0772登录后复制登录后复制 In [ ]
# To test 2:# python src/main.py test_withlabel --model-path-base [saved_model_directory]# since we don't offer labelled testset, you may not run the following command unless you create ~/data/test.txt by yourself#### !cd work; python src/main.py test_withlabel --model-path-base model-0/model_dev\=0.0772登录后复制
One more thing
You can continue to improve on the basis of the baseline model, or use a brand new model. You can try to tune from the following aspects:
Preprocess input data to extract more features. The baseline model uses the RNA secondary structure predicted from LinearFold as parts of features. You can try to add more features with other secondary structure prediction software (Vienna RNAfold, RNAstructure, CONTRAfold, etc.)More complex embeddingUse pre-trained models such as Elmo and Bert in the Embedding layerOptimize network structure and parametersHidden layer size selection-width and number of layersTry complex network architectureTry regularization, dropout, etc. to avoid overfittingOptimize hyperparameters such as learning rateChoose a good loss functionTry different optimizers