DGL.heterograph创建错误

忘是亡心i 2022-12-04 01:19 60阅读 0赞

raise DGLError("dgl.heterograph no longer supports graph construction from a SciPy "

data = getM4Ldata(data_file_path)
        data_dict = {
                ('Gene', 'bag-labled', 'Disease') : data['GvsD'],
               ('Gene', 'bag-instance', 'Isofrom') :data['GvsI'],
               ('Gene', 'Gene-LncRNA', 'LncRNA') : data['GvsL'],
               ('Gene', 'Gene-MiRNA', 'MiRNA') : data['GvsM']
                }
        G    = dgl.heterograph(data_dict)

在使用dgl.heterograph()方法创建异质信息网络时，报各种错误。可以进入官网查看参数[信息][Link 1]（[https://docs.dgl.ai/en/latest/generated/dgl.heterograph.html\#dgl.heterograph][Link 1]）。

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM5NDYzMTc1_size_16_color_FFFFFF_t_70][]

发现传入参数已经变换了，只能传入一维的tensor, 我的dgl版本是0.5.1，若想使用以前方式，重新下载，改为0.4.1以前即可

那么使用邻接矩阵创建异质图呢？

将numpy数据转为scipy数据，然后就可以直接构建异质图了

import dgl
    import dgl.function as fn
    import torch as th
    import torch.nn as nn
    import torch.nn.functional as F
    from dgl import DGLGraph
    import scipy.io
    import scipy.sparse as sp
    
    
    
    data_file_path = r'.\data\IsofromDataForDeepM4L.mat'
    gcn_msg = fn.copy_src(src='h', out='m')
    gcn_reduce = fn.sum(msg='m', out='h')
    
    
    def getM4Ldata(data_file_path):
        data = scipy.io.loadmat(data_file_path)
        return data
    data = getM4Ldata(data_file_path)
    #print(list(data.keys()))
    #['__header__', '__version__', '__globals__', 'GvsD', 'GvsI', 'GvsL', 'GvsM']
    
    def build_Up_HINM4L(data_file_path):
        data = getM4Ldata(data_file_path)
        data_dict = {
                ('Gene', 'bag-labled', 'Disease') :  sp.csr_matrix(data['GvsD']),
               ('Gene', 'bag-instance', 'Isofrom') : sp.csr_matrix(data['GvsI']),
               ('Gene', 'Gene-LncRNA', 'LncRNA') :  sp.csr_matrix(data['GvsL']),
               ('Gene', 'Gene-MiRNA', 'MiRNA') :  sp.csr_matrix(data['GvsM'])
                }
        G    = dgl.heterograph(data_dict)
        '''
        data_dict = {
                ('user', 'follows', 'user'): (torch.tensor([0, 1]), torch.tensor([1, 2])),
                ('user', 'follows', 'topic'): (torch.tensor([1, 1]), torch.tensor([1, 2])),
                ('user', 'plays', 'game'): (torch.tensor([0, 3]), torch.tensor([3, 4]))
                }
        da=dgl.heterograph(data_dict)
        '''
        return G
    Graph = build_Up_HINM4L(data_file_path)
    
    print(Graph)

其中data为cell数据，具体看下图（里面包含多个邻接矩阵数据）：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM5NDYzMTc1_size_16_color_FFFFFF_t_70 1][]

输出结果是：

![20200904202903340.png][]

[Link 1]: https://docs.dgl.ai/en/latest/generated/dgl.heterograph.html#dgl.heterograph
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM5NDYzMTc1_size_16_color_FFFFFF_t_70]: /images/20221123/b0d0e23b2afd4aa3801dfd81f90dbca9.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM5NDYzMTc1_size_16_color_FFFFFF_t_70 1]: /images/20221123/9b5ed94ea8884423b8d0024dab4e5e26.png
[20200904202903340.png]: /images/20221123/9ab0ac314e1641bf9157eabe2fa21e25.png