notebook.community

Edit and run 





In [1]:



    


import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import datetime
import os
# from small_script.myFunctions import *



%matplotlib inline
%load_ext autoreload
%autoreload 2



    











In [2]:



    


plt.rcParams['figure.figsize'] = [16.18033, 10]    #golden ratio
plt.rcParams['figure.facecolor'] = 'w'
plt.rcParams['figure.dpi'] = 100



    











In [6]:



    


a = pd.read_csv("/Users/weilu/Research/server/may_2019/database/Q_original_1r69", index_col=0)



    











In [79]:



    


b = a.query("Rank < 6000").sample(200)



    











In [81]:



    


all_movies = {}
for i in b["Run"].unique():
    with open(f"/Users/weilu/Research/server/may_2019/database/original_1r69_{i}/movie.pdb") as f:
        movie = f.readlines()
    all_movies[i] = movie
size = 0
for line in movie:
    size += 1
    if line == "END\n":
        break
print(size)



    


















    






314

















In [85]:



    


def getStructures(x, all_movies):
    index = int(x["index"])+1
    run = int(x["Run"])

    start = index * size
    end = (index + 1) * size
    f = io.StringIO("".join(all_movies[run][start:end]))
    return parser.get_structure(f"{index}", f)



    











In [92]:



    


b.iloc[0]



    


















    
Out[92]:








index                                                      923
 Qw                                                   0.535499
Run                                                          7
Rank                                                    2295.5
structure    (((<Residue SER het=  resseq=1 icode= >, <Resi...
Name: 7923, dtype: object

















In [91]:



    


xx = b.iloc[0]["structure"]



    











In [95]:



    


from Bio.PDB.PDBIO import PDBIO
out = PDBIO()
out.set_structure(xx)
out.save('/Users/weilu/Research/server/may_2019/out.pdb')



    











In [104]:



    


b.to_pickle("/Users/weilu/Research/server/may_2019/out.pkl")



    











In [110]:



    


c = pd.read_pickle("/Users/weilu/Research/server/may_2019/out.pkl")



    











In [ ]:



    


b["structure"] = b



    











In [114]:



    


b["structure"] = b.apply(getStructures, all_movies=all_movies, axis=1)



    











In [115]:



    


b.equals(c)



    


















    
Out[115]:








True

















In [13]:



    


314314/1001



    


















    
Out[13]:








314.0

















In [140]:



    


b



    


















    
Out[140]:











  
    

      
      index
      Qw
      Run
      Rank
      structure
      Qw
    

  
  
    

      7923
      923
      0.535499
      7
      2295.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.535
    

    

      1960
      960
      0.464596
      1
      1185.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.465
    

    

      10991
      991
      0.646153
      10
      255.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.646
    

    

      29946
      946
      0.556851
      29
      1605.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.557
    

    

      4954
      954
      0.629374
      4
      1365.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.629
    

    

      10982
      982
      0.644678
      10
      525.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.645
    

    

      13885
      885
      0.685323
      13
      3435.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.685
    

    

      6846
      846
      0.551446
      6
      4605.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.551
    

    

      25955
      955
      0.714764
      25
      1335.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.715
    

    

      21802
      802
      0.517790
      21
      5925.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.518
    

    

      14839
      839
      0.514815
      14
      4815.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.515
    

    

      29838
      838
      0.615635
      29
      4845.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.616
    

    

      2837
      837
      0.741397
      2
      4875.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.741
    

    

      15924
      924
      0.532348
      15
      2265.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.532
    

    

      14822
      822
      0.507551
      14
      5325.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.508
    

    

      28871
      871
      0.566710
      28
      3855.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.567
    

    

      24943
      943
      0.733249
      24
      1695.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.733
    

    

      17993
      993
      0.638778
      17
      195.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.639
    

    

      11821
      821
      0.518540
      11
      5355.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.519
    

    

      8901
      901
      0.602135
      8
      2955.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.602
    

    

      20878
      878
      0.475997
      20
      3645.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.476
    

    

      24911
      911
      0.681996
      24
      2655.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.682
    

    

      10996
      996
      0.657030
      10
      105.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.657
    

    

      4844
      844
      0.703470
      4
      4665.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.703
    

    

      20801
      801
      0.648037
      20
      5955.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.648
    

    

      5846
      846
      0.497530
      5
      4605.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.498
    

    

      11877
      877
      0.522874
      11
      3675.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.523
    

    

      9898
      898
      0.610686
      9
      3045.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.611
    

    

      25897
      897
      0.619894
      25
      3075.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.620
    

    

      10825
      825
      0.562865
      10
      5235.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.563
    

    

      ...
      ...
      ...
      ...
      ...
      ...
      ...
    

    

      6899
      899
      0.483526
      6
      3015.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.484
    

    

      4962
      962
      0.588982
      4
      1125.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.589
    

    

      16814
      814
      0.505825
      16
      5565.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.506
    

    

      17898
      898
      0.702807
      17
      3045.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.703
    

    

      2880
      880
      0.553256
      2
      3585.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.553
    

    

      1868
      868
      0.419378
      1
      3945.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.419
    

    

      18865
      865
      0.513220
      18
      4035.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.513
    

    

      15915
      915
      0.571546
      15
      2535.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.572
    

    

      22844
      844
      0.602857
      22
      4665.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.603
    

    

      11900
      900
      0.485654
      11
      2985.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.486
    

    

      7835
      835
      0.735204
      7
      4935.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.735
    

    

      21856
      856
      0.564427
      21
      4305.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.564
    

    

      29822
      822
      0.612041
      29
      5325.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.612
    

    

      8938
      938
      0.570013
      8
      1845.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.570
    

    

      12851
      851
      0.468245
      12
      4455.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.468
    

    

      18866
      866
      0.497601
      18
      4005.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.498
    

    

      6854
      854
      0.502701
      6
      4365.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.503
    

    

      15885
      885
      0.529569
      15
      3435.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.530
    

    

      19847
      847
      0.696022
      19
      4575.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.696
    

    

      934
      934
      0.537232
      0
      1965.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.537
    

    

      10947
      947
      0.651820
      10
      1575.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.652
    

    

      20869
      869
      0.470937
      20
      3915.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.471
    

    

      25919
      919
      0.626449
      25
      2415.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.626
    

    

      6838
      838
      0.513957
      6
      4845.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.514
    

    

      22883
      883
      0.549574
      22
      3495.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.550
    

    

      15937
      937
      0.520374
      15
      1875.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.520
    

    

      9960
      960
      0.578464
      9
      1185.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.578
    

    

      20803
      803
      0.673201
      20
      5895.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.673
    

    

      18826
      826
      0.677789
      18
      5205.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.678
    

    

      24885
      885
      0.558958
      24
      3435.5
      (((<Residue SER het=  resseq=1 icode= >, <Resi...
      0.559
    

  



200 rows × 6 columns



















In [143]:



    


c["structure"].tolist()



    


















    
Out[143]:








[<Structure id=924>,
 <Structure id=961>,
 <Structure id=992>,
 <Structure id=947>,
 <Structure id=955>,
 <Structure id=983>,
 <Structure id=886>,
 <Structure id=847>,
 <Structure id=956>,
 <Structure id=803>,
 <Structure id=840>,
 <Structure id=839>,
 <Structure id=838>,
 <Structure id=925>,
 <Structure id=823>,
 <Structure id=872>,
 <Structure id=944>,
 <Structure id=994>,
 <Structure id=822>,
 <Structure id=902>,
 <Structure id=879>,
 <Structure id=912>,
 <Structure id=997>,
 <Structure id=845>,
 <Structure id=802>,
 <Structure id=847>,
 <Structure id=878>,
 <Structure id=899>,
 <Structure id=898>,
 <Structure id=826>,
 <Structure id=853>,
 <Structure id=896>,
 <Structure id=868>,
 <Structure id=994>,
 <Structure id=895>,
 <Structure id=907>,
 <Structure id=918>,
 <Structure id=982>,
 <Structure id=894>,
 <Structure id=984>,
 <Structure id=846>,
 <Structure id=980>,
 <Structure id=869>,
 <Structure id=986>,
 <Structure id=810>,
 <Structure id=817>,
 <Structure id=882>,
 <Structure id=973>,
 <Structure id=980>,
 <Structure id=887>,
 <Structure id=936>,
 <Structure id=974>,
 <Structure id=852>,
 <Structure id=925>,
 <Structure id=842>,
 <Structure id=906>,
 <Structure id=875>,
 <Structure id=989>,
 <Structure id=824>,
 <Structure id=893>,
 <Structure id=898>,
 <Structure id=825>,
 <Structure id=884>,
 <Structure id=890>,
 <Structure id=833>,
 <Structure id=928>,
 <Structure id=900>,
 <Structure id=993>,
 <Structure id=892>,
 <Structure id=894>,
 <Structure id=828>,
 <Structure id=953>,
 <Structure id=998>,
 <Structure id=966>,
 <Structure id=921>,
 <Structure id=974>,
 <Structure id=952>,
 <Structure id=893>,
 <Structure id=823>,
 <Structure id=915>,
 <Structure id=918>,
 <Structure id=885>,
 <Structure id=882>,
 <Structure id=943>,
 <Structure id=975>,
 <Structure id=981>,
 <Structure id=926>,
 <Structure id=968>,
 <Structure id=970>,
 <Structure id=950>,
 <Structure id=997>,
 <Structure id=956>,
 <Structure id=820>,
 <Structure id=951>,
 <Structure id=888>,
 <Structure id=940>,
 <Structure id=913>,
 <Structure id=921>,
 <Structure id=929>,
 <Structure id=810>,
 <Structure id=992>,
 <Structure id=919>,
 <Structure id=826>,
 <Structure id=987>,
 <Structure id=972>,
 <Structure id=855>,
 <Structure id=962>,
 <Structure id=883>,
 <Structure id=835>,
 <Structure id=914>,
 <Structure id=969>,
 <Structure id=891>,
 <Structure id=887>,
 <Structure id=931>,
 <Structure id=866>,
 <Structure id=996>,
 <Structure id=997>,
 <Structure id=824>,
 <Structure id=969>,
 <Structure id=883>,
 <Structure id=840>,
 <Structure id=936>,
 <Structure id=843>,
 <Structure id=862>,
 <Structure id=820>,
 <Structure id=900>,
 <Structure id=900>,
 <Structure id=947>,
 <Structure id=995>,
 <Structure id=858>,
 <Structure id=861>,
 <Structure id=865>,
 <Structure id=965>,
 <Structure id=956>,
 <Structure id=974>,
 <Structure id=923>,
 <Structure id=880>,
 <Structure id=802>,
 <Structure id=938>,
 <Structure id=959>,
 <Structure id=910>,
 <Structure id=882>,
 <Structure id=835>,
 <Structure id=991>,
 <Structure id=934>,
 <Structure id=833>,
 <Structure id=859>,
 <Structure id=868>,
 <Structure id=811>,
 <Structure id=829>,
 <Structure id=864>,
 <Structure id=808>,
 <Structure id=966>,
 <Structure id=941>,
 <Structure id=999>,
 <Structure id=837>,
 <Structure id=892>,
 <Structure id=855>,
 <Structure id=916>,
 <Structure id=930>,
 <Structure id=993>,
 <Structure id=851>,
 <Structure id=917>,
 <Structure id=851>,
 <Structure id=853>,
 <Structure id=860>,
 <Structure id=883>,
 <Structure id=919>,
 <Structure id=810>,
 <Structure id=881>,
 <Structure id=900>,
 <Structure id=963>,
 <Structure id=815>,
 <Structure id=899>,
 <Structure id=881>,
 <Structure id=869>,
 <Structure id=866>,
 <Structure id=916>,
 <Structure id=845>,
 <Structure id=901>,
 <Structure id=836>,
 <Structure id=857>,
 <Structure id=823>,
 <Structure id=939>,
 <Structure id=852>,
 <Structure id=867>,
 <Structure id=855>,
 <Structure id=886>,
 <Structure id=848>,
 <Structure id=935>,
 <Structure id=948>,
 <Structure id=870>,
 <Structure id=920>,
 <Structure id=839>,
 <Structure id=884>,
 <Structure id=938>,
 <Structure id=961>,
 <Structure id=804>,
 <Structure id=827>,
 <Structure id=886>]

















In [14]:



    


with open("/Users/weilu/Research/server/may_2019/database/original_1r69_0/movie.pdb") as f:
    movie = f.readlines()



    











In [55]:



    






    











In [56]:



    


size



    


















    
Out[56]:








314

















In [53]:



    


movie[-1]



    


















    
Out[53]:








'END\n'

















In [15]:



    


len(movie)



    


















    
Out[15]:








314314

















In [33]:



    


f = io.StringIO("".join(movie[:314]))



    











In [34]:



    


from Bio.PDB.PDBParser import PDBParser
parser = PDBParser()
x = parser.get_structure("test", f)



    











In [134]:



    


a1, a2 = "origianl_frag__1r69_2".split("__")



    











In [135]:



    


a1



    


















    
Out[135]:








'origianl_frag'

















In [136]:



    


a2



    


















    
Out[136]:








'1r69_2'

















In [19]:



    


import io



    











In [17]:



    


import tempfile



    











In [ ]:



    






    











In [ ]:



    


tempfile.TemporaryFile



    











In [ ]:



    


temp_file = tempfile.NamedTemproaryFile(delete = False)



    











In [ ]:



    


from Bio.PDB.PDBParser import PDBParser
import numpy , tempfile ,os , re

models_re = re.compile("END")
pdb_re = re.compile(r"MODEL(.*?)ENDMDL", re.DOTALL)

def PDB_parse(pdb_file_handle):
    model_pos = []
    models = []
    k = open(pdb_file_handle,"r").read()
    for i in models_re.finditer(k):
        model_pos.append(i.start())
    for i in model_pos:
        models.append(pdb_re.search(k,i).group())
    return models

array_all_structure = []

for i in PDB_parse(pdb_file_handle):
    temp_file = tempfile.NamedTemproaryFile(delete = False)
    temp_file.write(i)
    temp_file.close
    structure = parser.get_structure("1fqy", temp_file.name)
    os.remove(temp_file.name)
    model = structure[0]
    chain = model["A"]
    S1coor = numpy.zeros(shape=(226, 3))
    i = 0
    for residue1 in chain:
       resnum = residue1.get_id()[1]
       atom1 = residue1['CA']
       S1coor[i] = atom1.get_coord()
       i = i + 1
       array_all_structure.append(i)

Content source: luwei0917/awsemmd_script

Similar notebooks:

notebook.community | gallery | about