libsonata 0.1.28

NodeStorage

>>> import libsonata

>>> nodes = libsonata.NodeStorage('path/to/H5/file')

# list populations
>>> nodes.population_names

# open population
>>> population = nodes.open_population(<name>)

NodePopulation

# total number of nodes in the population
>>> population.size

# attribute names
>>> population.attribute_names

# get attribute value for single node, say 42
>>> population.get_attribute('mtype', 42)

# ...or Selection of nodes (see below) => returns NumPy array with corresponding values
>>> selection = libsonata.Selection(values=[1, 5, 9, 42])  # nodes 1, 5, 9, 42
>>> mtypes = population.get_attribute('mtype', selection)
>>> list(zip(selection.flatten(), mtypes))
[(1, u'mtype_of_1'), (5, u'mtype_of_5'), (9, u'mtype_of_9'), (42, u'mtype_of_42')]

选择

元素ID列表（可以是node_id或edge_id），相邻ID被分组以提高HDF5文件访问效率。例如，{1, 2, 3, 5}序列变为{[1, 4), [5, 6)}

选择可以从以下实例化

• 标量值序列（也适用于NumPy数组）

• 对序列（解释为上述范围，也适用于N x 2 NumPy数组）

EdgePopulation 连接查询（见下文）也返回 选择。

>>> selection = libsonata.Selection([1, 2, 3, 5])
>>> selection.ranges
[(1, 4), (5, 6)]

>>> selection = libsonata.Selection([(1, 4), (5, 6)])
>>> selection.flatten()
[1, 2, 3, 5]
>>> selection.flat_size
4
>>> bool(selection)
True

节点集合

libsonata 可以与节点集合概念协同工作，如以下所述： SONATA指南：节点集合文件 这允许为细胞组定义名称，并提供查询它们的方法。libsonata 还允许使用扩展表达式，例如正则表达式和浮点数测试，如以下所述： SONATA扩展：节点集合

# load a node set JSON file
>>> node_sets = libsonata.NodeSets.from_file('node_sets.json')

# list node sets
>>> node_sets.names
{'L6_UPC', 'Layer1', 'Layer2', 'Layer3', ....}

# get the selection of nodes that match in population
>>> selection = node_sets.materialize('Layer1', population)

# node sets can also be loaded from a JSON string
>>> node_sets_manual = libsonata.NodeSets(json.dumps({"SLM_PPA_and_SP_PC": {"mtype": ["SLM_PPA", "SP_PC"]}}))
>>> node_sets_manual.names
{'SLM_PPA_and_SP_PC'}

边存储

EdgeStorage 的节点处理与 NodeStorage 类似

>>> edges = libsonata.EdgeStorage('path/to/H5/file')

# list populations
>>> edges.population_names

# open population
>>> population = edges.open_population(<name>)

边人口

# total number of edges in the population
>>> population.size

# attribute names
>>> population.attribute_names

# get attribute value for single edge, say 123
>>> population.get_attribute('delay', 123)

# ...or Selection of edges => returns NumPy array with corresponding values
>>> selection = libsonata.Selection([1, 5, 9])
>>> population.get_attribute('delay', selection) # returns delays for edges 1, 5, 9

…提供额外的查询连接的方法，结果为可以像上面一样应用的选择

# get source / target node ID for the 42nd edge:
>>> population.source_node(42)
>>> population.target_node(42)

# query connectivity (result is Selection object)
>>> selection_to_1 = population.afferent_edges(1)  # all edges with target node_id 1
>>> population.target_nodes(selection_to_1)  # since selection only contains edges
                                             # targeting node_id 1 the result will be a
                                             # numpy array of all 1's
>>> selection_from_2 = population.efferent_edges(2)  # all edges sourced from node_id 2
>>> selection = population.connecting_edges(2, 1)  # this selection is all edges from
                                                   # node_id 2 to node_id 1

# ...or their vectorized analogues
>>> selection = population.afferent_edges([1, 2, 3])
>>> selection = population.efferent_edges([1, 2, 3])
>>> selection = population.connecting_edges([1, 2, 3], [4, 5, 6])

尖峰读取器

>>> import libsonata

>>> spikes = libsonata.SpikeReader('path/to/H5/file')

# list populations
>>> spikes.get_population_names()

# open population
>>> population = spikes['<name>']

尖峰人口

# get all spikes [(node_id, timestep)]
>>> population.get()
[(5, 0.1), (2, 0.2), (3, 0.3), (2, 0.7), (3, 1.3)]

# get all spikes betwen tstart and tstop
>>> population.get(tstart=0.2, tstop=1.0)
[(2, 0.2), (3, 0.3), (2, 0.7)]

# get spikes attribute sorting (by_time, by_id, none)
>>> population.sorting
'by_time'

Pandas can be used to create a dataframe and get a better representation of the data

>>> import pandas

data = population.get()
df = pandas.DataFrame(data=data, columns=['ids', 'times']).set_index('times')
print(df)
       ids
times
0.1      5
0.2      2
0.3      3
0.7      2
1.3      3

细胞体报告读取器

>>> somas = libsonata.SomaReportReader('path/to/H5/file')

# list populations
>>> somas.get_population_names()

# open population
>>> population_somas = somas['<name>']

细胞体报告人口

# get times (tstart, tstop, dt)
>>> population_somas.times
(0.0, 1.0, 0.1)

# get unit attributes
>>> population_somas.time_units
'ms'
>>> population_somas.data_units
'mV'

# node_ids sorted?
>>> population_somas.sorted
True

# get a list of all node ids in the selected population
>>> population_somas.get_node_ids()
[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20]

# get the DataFrame of the node_id values for the timesteps between tstart and tstop
>>> data_frame = population_somas.get(node_ids=[13, 14], tstart=0.8, tstop=1.0)

# get the data values
>>> data_frame.data
[[13.8, 14.8], [13.9, 14.9]]

# get the list of timesteps
>>> data_frame.times
[0.8, 0.9]

# get the list of node ids
>>> data_frame.ids
[13, 14]

再次使用pandas，可以创建一个使用数据、id和时间的列表的dataframe

>>> import pandas

df = pandas.DataFrame(data_frame.data, columns=data_frame.ids, index=data_frame.times)
print(df)
       13    14
0.8  13.8  14.8
0.9  13.9  14.9

元素报告读取器

>>> elements = libsonata.ElementReportReader('path/to/H5/file')

# list populations
>>> elements.get_population_names()

# open population
>>> population_elements = elements['<name>']

元素报告人口

# get times (tstart, tstop, dt)
>>> population_elements.times
(0.0, 4.0, 0.2)

>>> population_elements.get_node_ids()
[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20]

# get the DataFrame of the node_id values for the timesteps between tstart and tstop
>>> data_frame = population_elements.get(node_ids=[13, 14], tstart=0.8, tstop=1.0)

# get the data values (list of list of floats with data[time_index][element_index])
>>> data_frame.data
[[46.0, 46.1, 46.2, 46.3, 46.4, 46.5, 46.6, 46.7, 46.8, 46.9], [56.0, 56.1, 56.2, 56.3, 56.4, 56.5, 56.6, 56.7, 56.8, 56.9]]

# get the list of timesteps
>>> data_frame.times
[0.8, 1.0]

# get the list of (node id, element_id)
>>> data_frame.ids
[(13, 30), (13, 30), (13, 31), (13, 31), (13, 32), (14, 32), (14, 33), (14, 33), (14, 34), (14, 34)]

与尖峰和细胞体报告一样，可以使用pandas更好地表示数据

>>> import pandas

df = pandas.DataFrame(data_frame.data, columns=pandas.MultiIndex.from_tuples(data_frame.ids), index=data_frame.times)
print(df)
       13                            14
       30    30    31    31    32    32    33    33    34    34
0.8  46.0  46.1  46.2  46.3  46.4  46.5  46.6  46.7  46.8  46.9
1.0  56.0  56.1  56.2  56.3  56.4  56.5  56.6  56.7  56.8  56.9

对于大数据集，使用numpy数组可以大大提高性能

>>> import numpy

np_data = numpy.asarray(data_frame.data)
np_ids = numpy.asarray(data_frame.ids).T
np_times = numpy.asarray(data_frame.times)

df = pandas.DataFrame(np_data, columns=pandas.MultiIndex.from_arrays(np_ids), index=np_times)