1

我有一个制表符分隔的文件,其中包含区域和在这些区域中找到的各个生物实体(我检查了 67 个,因此您说每个区域都检查了这 67 个实体的存在或不存在及其频率)。

我有所有这些表格格式的数据。

下面给出了一个示例数据

Region  ATF3    BCL3    BCLAF1  BDP1    BRF1    BRF2    Brg1    CCNT2   CEBPB   CHD2    CTCF    CTCFL   E2F6    ELF1
chr1:109102470:109102970    0   0   1   0   0   0   0   1   0   0   4   1   4   1
chr1:110526886:110527386    0   0   0   0   0   0   0   1   1   0   4   1   0   1
chr1:115300671:115301171    0   0   1   0   0   0   0   0   1   1   4   1   1   1
chr1:115323308:115323808    0   0   0   0   0   0   0   1   0   0   2   1   1   0
chr1:11795641:11796141  1   0   0   0   0   0   0   1   2   0   0   0   1   0
chr1:118148103:118148603    0   0   0   0   0   0   0   1   0   0   0   0   0   1
chr1:150521397:150521897    0   0   0   0   0   0   0   2   2   0   6   2   4   0
chr1:150601609:150602109    0   0   0   0   0   0   0   0   3   2   0   0   1   0
chr1:150602098:150602598    0   0   0   0   0   0   0   0   1   1   0   0   0   0
chr1:151119140:151119640    0   0   0   0   0   0   0   1   0   0   0   0   1   0
chr1:151128604:151129104    0   0   0   0   0   0   0   0   0   0   3   0   0   0
chr1:153517729:153518229    0   0   0   0   0   0   0   0   0   0   0   0   0   0
chr1:153962738:153963238    0   0   0   0   0   0   0   1   1   0   0   0   0   1
chr1:154155682:154156182    0   0   0   0   0   0   0   1   0   0   0   0   1   1
chr1:154155725:154156225    0   0   0   0   0   0   0   1   0   0   0   0   1   1
chr1:154192154:154192654    0   0   0   0   0   0   0   0   0   0   0   0   0   0
chr1:154192824:154193324    1   0   0   0   0   0   0   1   0   1   0   0   1   1
chr1:154192943:154193443    1   0   0   0   0   0   0   1   0   2   0   0   1   1
chr1:154193273:154193773    1   0   0   0   0   0   0   1   0   2   0   0   2   1
chr1:154193313:154193813    0   0   0   0   0   0   0   1   0   2   0   0   2   1
chr1:155904188:155904688    0   0   0   0   0   0   0   1   0   0   0   0   1   1
chr1:155947966:155948466    0   0   0   0   0   0   0   1   0   0   3   0   0   1
chr1:155948336:155948836    0   0   0   0   0   0   0   1   0   0   5   1   0   1
chr1:156023516:156024016    0   0   0   0   0   0   0   1   0   1   4   1   1   1
chr1:156024016:156024516    0   1   1   0   0   0   0   0   0   2   0   0   1   1
chr1:156163229:156163729    0   0   0   0   0   0   0   0   0   0   2   0   0   1
chr1:160990902:160991402    0   0   0   0   0   0   0   0   0   1   0   0   1   2
chr1:160991133:160991633    0   0   0   0   0   0   0   0   0   1   0   0   1   2
chr1:161474704:161475204    0   0   0   0   0   0   0   0   0   0   0   0   0   0
chr1:161509530:161510030    0   0   1   1   1   0   0   0   1   0   1   0   0   1
chr1:161590964:161591464    0   0   0   1   1   0   0   0   0   0   0   0   0   0
chr1:169075446:169075946    0   0   0   0   0   0   0   2   0   0   4   0   3   0
chr1:17053279:17053779  0   0   0   1   0   0   0   0   0   1   0   0   0   0
chr1:1709909:1710409    0   0   0   0   0   0   0   2   0   1   0   0   3   1
chr1:1710297:1710797    0   0   0   0   0   0   0   0   0   1   6   0   1   1

现在我怎么能把它放在从颜色代码浅红色到深红色的热图中(取决于频率和白色,如果没有的话)?

还有其他更好的方法来表示这种类型的数据吗?

4

2 回答 2

4

使用Matplotlib

import pylab as plt
import numpy as np

data = np.loadtxt("14318737.txt", skiprows=1, converters={0:lambda x: 0})
plot_data = np.ma.masked_equal(data[:,1:], 0)

plt.imshow(plot_data, cmap=plt.cm.get_cmap("Reds"), interpolation="nearest")
plt.colorbar()

plt.show()

我忽略了第一行和第一列(如果您需要它们作为标签,我们需要更改它)。对于剩余的数据,所有零值都被屏蔽(因此它们在图中显示为白色),然后将这些数据绘制为彩色图。

imshow有一堆其他参数来控制结果,例如原点(下/上),方面(自动/相等/some_ratio)。

你写的是地区——你是指地理区域吗?然后你可能想看看Matplotlib 的 Basemap Toolkit来创建颜色编码的地图。

编辑

新要求,新范例

import pylab as plt
import numpy as np

fn = "14318737.txt"
with open(fn, "r") as f:
    labels = f.readline().rstrip("\n").split()[1:]
data = np.loadtxt(fn, skiprows=1, converters={0:lambda x: 0})
plot_data = np.ma.masked_equal(data[:,1:], 0)

plt.subplots_adjust(left=0.1, bottom=0.15, right=0.99, top=0.95)
plt.imshow(plot_data, cmap=plt.cm.get_cmap("Reds"), interpolation="nearest", aspect = "auto")
plt.xticks(range(len(labels)), labels, rotation=90, va="top", ha="center")
plt.colorbar()

plt.show()

现在我首先从第一行读取标签。aspect我在imshow-call 中添加了关键字参数。我为每个因素创建标签。

此外,我使用subplots_adjust. 您可以使用这些参数,直到它们满足您的需求。

现在的结果是: 结果热图

如果您想要 y 轴的其他刻度,请使用plt.yticks,就像xticks在我的示例中一样。

于 2013-01-14T13:16:15.297 回答
1

由于对我的其他答案的评论,OP 有另一个关于搜索 2d 集群的问题。这是一些答案。

取自我的图书馆eegpy,我使用了一种方法 find_clusters。它遍历二维数组,找到高于/低于给定阈值的所有集群。

这是我的代码:

import pylab as plt
import numpy as np
from Queue import Queue


def find_clusters(ar,thres,cmp_type="greater"):
    """For a given 2d-array (test statistic), find all clusters which
are above/below a certain threshold.
"""
    if not cmp_type in ["lower","greater","abs_greater"]:
        raise ValueError("cmp_type must be in [\"lower\",\"greater\",\"abs_greater\"]")
    clusters = []
    if cmp_type=="lower":
        ar_in = (ar<thres).astype(np.bool)
    elif cmp_type=="greater":
        ar_in = (ar>thres).astype(np.bool)
    else: #cmp_type=="abs_greater":
        ar_in = (abs(ar)>thres).astype(np.bool)

    already_visited = np.zeros(ar_in.shape,np.bool)
    for i_s in range(ar_in.shape[0]): #i_s wie i_sample
        for i_f in range(ar_in.shape[1]):
            if not already_visited[i_s,i_f]:
                if ar_in[i_s,i_f]:
                    #print "Anzahl cluster:", len(clusters)
                    mask = np.zeros(ar_in.shape,np.bool)
                    check_queue = Queue()
                    check_queue.put((i_s,i_f))
                    while not check_queue.empty():
                        pos_x,pos_y = check_queue.get()
                        if not already_visited[pos_x,pos_y]:
                            #print pos_x,pos_y
                            already_visited[pos_x,pos_y] = True
                            if ar_in[pos_x,pos_y]:
                                mask[pos_x,pos_y] = True
                                for coords in [(pos_x-1,pos_y),(pos_x+1,pos_y),(pos_x,pos_y-1),(pos_x,pos_y+1)]: #Direct Neighbors
                                    if 0<=coords[0]<ar_in.shape[0] and 0<=coords[1]<ar_in.shape[1]:
                                        check_queue.put(coords)
                    clusters.append(mask)
    return clusters

fn = "14318737.txt"
with open(fn, "r") as f:
    labels = f.readline().rstrip("\n").split()[1:]
data = np.loadtxt(fn, skiprows=1, converters={0:lambda x: 0})

clusters = find_clusters(data, 0, "greater")

plot_data = np.ma.masked_equal(data[:,1:], 0)

plt.subplots_adjust(left=0.1, bottom=0.15, right=0.99, top=0.95)
plt.imshow(plot_data, cmap=plt.cm.get_cmap("Reds"), interpolation="nearest", aspect = "auto", 
           vmin=0, extent=[0.5,plot_data.shape[1]+0.5, plot_data.shape[0] - 0.5, -0.5])
plt.colorbar()

for cl in clusters:
    plt.contour(cl.astype(np.int),[0.5], colors="k", lw=2)
plt.xticks(np.arange(1, len(labels)+2), labels, rotation=90, va="top", ha="center")


plt.show()

它给出了以下形式的图像:

在集群周围绘制轮廓

clusters是布尔二维数组 (True / False) 的列表。每个数组代表一个簇,其中每个布尔值指示特定“点”是否是该簇的一部分。您可以在任何进一步的分析中使用它。

编辑

现在在集群上有更多的乐趣

import pylab as plt
import numpy as np
from Queue import Queue


def find_clusters(ar,thres,cmp_type="greater"):
    """For a given 2d-array (test statistic), find all clusters which
are above/below a certain threshold.
"""
    if not cmp_type in ["lower","greater","abs_greater"]:
        raise ValueError("cmp_type must be in [\"lower\",\"greater\",\"abs_greater\"]")
    clusters = []
    if cmp_type=="lower":
        ar_in = (ar<thres).astype(np.bool)
    elif cmp_type=="greater":
        ar_in = (ar>thres).astype(np.bool)
    else: #cmp_type=="abs_greater":
        ar_in = (abs(ar)>thres).astype(np.bool)

    already_visited = np.zeros(ar_in.shape,np.bool)
    for i_s in range(ar_in.shape[0]): #i_s wie i_sample
        for i_f in range(ar_in.shape[1]):
            if not already_visited[i_s,i_f]:
                if ar_in[i_s,i_f]:
                    #print "Anzahl cluster:", len(clusters)
                    mask = np.zeros(ar_in.shape,np.bool)
                    check_queue = Queue()
                    check_queue.put((i_s,i_f))
                    while not check_queue.empty():
                        pos_x,pos_y = check_queue.get()
                        if not already_visited[pos_x,pos_y]:
                            #print pos_x,pos_y
                            already_visited[pos_x,pos_y] = True
                            if ar_in[pos_x,pos_y]:
                                mask[pos_x,pos_y] = True
                                for coords in [(pos_x-1,pos_y),(pos_x+1,pos_y),(pos_x,pos_y-1),(pos_x,pos_y+1)]: #Direct Neighbors
                                    if 0<=coords[0]<ar_in.shape[0] and 0<=coords[1]<ar_in.shape[1]:
                                        check_queue.put(coords)
                    clusters.append(mask)
    return clusters

fn = "14318737.txt"
data = []
with open(fn, "r") as f:
    labels = f.readline().rstrip("\n").split()[1:]
    for line in f:
        data.append([int(v) for v in line.split()[1:]])
data = np.array(data) #np.loadtxt(fn, skiprows=1, usecols=range(1,15))#converters={0:lambda x: 0})

clusters = find_clusters(data, 0, "greater")
large_clusters = filter(lambda cl: cl.sum()>5, clusters) #Only take clusters with five or more items
large_clusters = sorted(large_clusters, key=lambda cl: -cl.sum())

plot_data = np.ma.masked_equal(data[:,:], 0)

plt.subplots_adjust(left=0.1, bottom=0.15, right=0.99, top=0.95)
plt.imshow(plot_data, cmap=plt.cm.get_cmap("Reds"), interpolation="nearest", aspect = "auto", 
           vmin=0, extent=[-0.5,plot_data.shape[1]-0.5, plot_data.shape[0] - 0.5, -0.5])
plt.colorbar()

for cl in large_clusters:
    plt.contour(cl.astype(np.int),[.5], colors="k", lw=2)
plt.xticks(np.arange(0, len(labels)+1), labels, rotation=90, va="top", ha="center")

print "Summary of all large clusters:\n"
print "#\tSize\tIn regions"
for i, cl in enumerate(large_clusters):
    print "%i\t%i\t" % (i, cl.sum()),
    regions_in_cluster = np.where(np.any(cl, axis=0))[0]
    min_region = labels[min(regions_in_cluster)]
    max_region = labels[max(regions_in_cluster)]
    print "%s to %s" % (min_region, max_region)

plt.xlim(-0.5,plot_data.shape[1]-0.5)

plt.show()

我过滤了所有包含五个以上点的集群。我只绘制这些。您也可以使用data每个集群内部的总和。然后我按它们的大小对这些大集群进行降序排序。

最后,我打印了所有大型集群的摘要,包括它们所在的所有集群的名称。 仅限大型集群

于 2013-05-08T07:19:57.573 回答