notebook.community

Edit and run



In [1]:

    
from lxml import etree
import pandas as pd



In [2]:

    
tree = etree.parse('biosample_result.xml')



In [3]:

    
# dict of all XPATHs we're interested in, relative to a given <BioSample>
BIOSAMPLE_XPATHS = {
    "biosample_id": ".//Id[@db='BioSample']", 
    "sra_id": ".//Id[@db='SRA']", 
    "title": ".//Title", 
    "paragraph": ".//Description/Comment/Paragraph",
    "tissue": ".//Attribute[@attribute_name='tissue']",
    "age": ".//Attribute[@attribute_name='age']",
    "cultivar": ".//Attribute[@attribute_name='cultivar']",
    "maize_cultivar": ".//Attribute[@attribute_name='maize cultivar']",
    "genotype": ".//Attribute[@attribute_name='genotype']",
    "ecotype": ".//Attribute[@attribute_name='ecotype']",
    "isolate": ".//Attribute[@attribute_name='isolate']",
    "dev_stage": ".//Attribute[@attribute_name='dev_stage']",
    "leaf_number": ".//Attribute[@attribute_name='leaf number']",
    "source_name": ".//Attribute[@attribute_name='source name']",
    "label": ".//Attribute[@attribute_name='label']"
}



In [4]:

    
def extract_biosample_values(biosample, xpaths):
    """Takes a <BioSample> element and a dict that maps
    from attribute names to XPATHs.
    Extracts all the text values from the XPATHs.
    Returns a dict mapping from the attribute names to
    attribute values.
    """
    result = {}
    for name, query in xpaths.items():
        xpath_result = biosample.xpath(query)
        if xpath_result:
            # XPATH queries always return a list of results,
            # but we're certain there's only one result.
            result[name] = xpath_result[0].text
        else: # If you remove the "else"-part,
              # the resulting table will contain NaNs instead of empty strings
            result[name] = ""
    return result



In [5]:

    
extracted_samples = []
for sample in tree.iter('BioSample'):
    extracted_sample = extract_biosample_values(sample, BIOSAMPLE_XPATHS)
    extracted_samples.append(extracted_sample)



In [6]:

    
samples_table = pd.DataFrame(extracted_samples)



In [7]:

    
samples_table









    Out[7]:







  
    
      
      age
      biosample_id
      cultivar
      dev_stage
      ecotype
      genotype
      isolate
      label
      leaf_number
      maize_cultivar
      paragraph
      source_name
      sra_id
      tissue
      title
    
  
  
    
      0
      9
      SAMEA5605513
      B73
      
      
      wild type genotype
      
      
      
      
      
      
      ERS3409710
      
      E-MTAB-7200_2:Sample 9
    
    
      1
      9
      SAMEA5605512
      B73
      
      
      wild type genotype
      
      
      
      
      
      
      ERS3409709
      
      E-MTAB-7200_2:Sample 8
    
    
      2
      9
      SAMEA5605511
      B73
      
      
      wild type genotype
      
      
      
      
      
      
      ERS3409708
      
      E-MTAB-7200_2:Sample 7
    
    
      3
      9
      SAMEA5605510
      B73
      
      
      wild type genotype
      
      
      
      
      
      
      ERS3409707
      
      E-MTAB-7200_2:Sample 6
    
    
      4
      9
      SAMEA5605509
      B73
      
      
      wild type genotype
      
      
      
      
      
      
      ERS3409706
      
      E-MTAB-7200_2:Sample 5
    
    
      5
      9
      SAMEA5605508
      B73
      
      
      wild type genotype
      
      
      
      
      
      
      ERS3409705
      
      E-MTAB-7200_2:Sample 4
    
    
      6
      9
      SAMEA5605507
      B73
      
      
      wild type genotype
      
      
      
      
      
      
      ERS3409704
      
      E-MTAB-7200_2:Sample 3
    
    
      7
      9
      SAMEA5605506
      B73
      
      
      wild type genotype
      
      
      
      
      
      
      ERS3409703
      
      E-MTAB-7200_2:Sample 2
    
    
      8
      9
      SAMEA5605505
      B73
      
      
      wild type genotype
      
      
      
      
      
      
      ERS3409702
      
      E-MTAB-7200_2:Sample 10
    
    
      9
      9
      SAMEA5605504
      B73
      
      
      wild type genotype
      
      
      
      
      
      
      ERS3409701
      
      E-MTAB-7200_2:Sample 1
    
    
      10
      
      SAMEA5229560
      
      
      
      
      
      
      
      
      
      
      ERS3037001
      
      B73
    
    
      11
      7 days
      SAMN03217742
      Zea mays
      
      B73
      
      
      
      
      
      In this study, temporal changes of gene and mi...
      
      SRS750388
      leaf
      Greening process of maize and rice leaves
    
    
      12
      13 days after sowing
      SAMN03100283
      B73
      
      
      
      
      
      
      
      
      
      SRS718868
      etiolated tissue
      Zea mays subsp. mays: B73 DNA
    
    
      13
      
      SAMN03031762
      B73
      
      
      lg1-R
      
      
      
      
      
      
      SRS700189
      P6 leaf primordia
      lg1-3
    
    
      14
      
      SAMN03031761
      B73
      
      
      lg1-R
      
      
      
      
      
      
      SRS700188
      P6 leaf primordia
      lg1-2
    
    
      15
      
      SAMN03031760
      B73
      
      
      lg1-R
      
      
      
      
      
      
      SRS700187
      P6 leaf primordia
      lg1-1
    
    
      16
      
      SAMN03031759
      B73
      
      
      wild type
      
      
      
      
      
      
      SRS700186
      P6 leaf primordia
      wtL-3
    
    
      17
      
      SAMN03031758
      B73
      
      
      wild type
      
      
      
      
      
      
      SRS700185
      P6 leaf primordia
      wtL-2
    
    
      18
      
      SAMN03031757
      B73
      
      
      wild type
      
      
      
      
      
      
      SRS700184
      P6 leaf primordia
      wtL-1
    
    
      19
      2 weeks old
      SAMN02996429
      B73
      
      
      
      
      
      
      
      
      
      SRS685657
      shoot meristem
      Plant sample from Zea mays subsp. mays
    
    
      20
      2 weeks old
      SAMN02996428
      B73
      
      
      
      
      
      
      
      
      
      SRS685656
      Shoot meristem
      Plant sample from Zea mays subsp. mays
    
    
      21
      2 weeks old
      SAMN02996427
      B73
      
      
      
      
      
      
      
      
      
      SRS685655
      laser microdissected shoot meristem
      Plant sample from Zea mays subsp. mays
    
    
      22
      2 weeks old
      SAMN02996426
      B73
      
      
      
      
      
      
      
      
      
      SRS685654
      Whole seedling
      Plant sample from Zea mays subsp. mays
    
    
      23
      2 weeks
      SAMN02996425
      B73
      
      
      
      
      
      
      
      
      
      SRS685653
      Whole seedling
      Plant sample from Zea mays subsp. mays
    
    
      24
      
      SAMN02401801
      
      
      
      
      
      
      
      
      
      
      SRS502117
      
      PARE library from Maize B73, pollen
    
    
      25
      
      SAMN02351542
      inbred line Mo17
      
      
      
      
      B73-stele
      
      
      RNA isolated from individual tissues of young ...
      
      SRS478255
      
      Zea mays, Mo17 4-day-old primary roots from st...
    
    
      26
      
      SAMN02203474
      
      
      
      
      
      
      
      
      
      
      SRS444558
      
      Zea Mays, B73 x Mo17 BSA tall
    
    
      27
      
      SAMN02203473
      
      
      
      
      
      
      
      
      
      
      SRS444557
      
      Zea Mays, B73 x Mo17 BSA short
    
    
      28
      
      SAMN02203472
      
      
      
      
      
      
      
      
      
      
      SRS444556
      
      Zea Mays, B73 x Mo17 BSA late
    
    
      29
      
      SAMN02203471
      
      
      
      
      
      
      
      
      
      
      SRS444555
      
      Zea Mays, B73 x Mo17 BSA early
    
    
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
    
    
      3858
      36
      SAMN12138014
      B73
      ZM2
      
      
      
      
      
      
      
      
      SRS5023549
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3859
      36
      SAMN12138013
      B73
      ZM3
      
      
      
      
      
      
      
      
      SRS5023547
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3860
      38
      SAMN12138012
      B73
      ZM3
      
      
      
      
      
      
      
      
      SRS5023546
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3861
      38
      SAMN12138011
      B73
      ZM3
      
      
      
      
      
      
      
      
      SRS5023545
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3862
      38
      SAMN12138010
      B73
      ZM4
      
      
      
      
      
      
      
      
      SRS5023544
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3863
      38
      SAMN12138009
      B73
      ZM3
      
      
      
      
      
      
      
      
      SRS5023543
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3864
      38
      SAMN12138008
      B73
      ZM3
      
      
      
      
      
      
      
      
      SRS5023542
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3865
      38
      SAMN12138007
      B73
      ZM2
      
      
      
      
      
      
      
      
      SRS5023541
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3866
      39
      SAMN12138006
      B73
      ZM3
      
      
      
      
      
      
      
      
      SRS5023539
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3867
      39
      SAMN12138005
      B73
      ZM4
      
      
      
      
      
      
      
      
      SRS5023540
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3868
      39
      SAMN12138004
      B73
      ZM3
      
      
      
      
      
      
      
      
      SRS5023550
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3869
      39
      SAMN12138003
      B73
      ZM4
      
      
      
      
      
      
      
      
      SRS5023551
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3870
      40
      SAMN12138002
      B73
      ZM4
      
      
      
      
      
      
      
      
      SRS5023563
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3871
      40
      SAMN12138001
      B73
      ZM3
      
      
      
      
      
      
      
      
      SRS5023565
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3872
      40
      SAMN12138000
      B73
      ZM4
      
      
      
      
      
      
      
      
      SRS5023562
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3873
      41
      SAMN12137999
      B73
      ZM5
      
      
      
      
      
      
      
      
      SRS5023564
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3874
      41
      SAMN12137998
      B73
      ZM5
      
      
      
      
      
      
      
      
      SRS5023560
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3875
      41
      SAMN12137997
      B73
      ZM4
      
      
      
      
      
      
      
      
      SRS5023561
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3876
      41
      SAMN12137996
      B73
      ZM4
      
      
      
      
      
      
      
      
      SRS5023558
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3877
      42
      SAMN12137995
      B73
      ZM5
      
      
      
      
      
      
      
      
      SRS5023559
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3878
      42
      SAMN12137994
      B73
      ZM5
      
      
      
      
      
      
      
      
      SRS5023576
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3879
      42
      SAMN12137993
      B73
      ZM5
      
      
      
      
      
      
      
      
      SRS5023577
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3880
      42
      SAMN12137992
      B73
      ZM5
      
      
      
      
      
      
      
      
      SRS5023574
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3881
      42
      SAMN12137991
      B73
      ZM4
      
      
      
      
      
      
      
      
      SRS5023575
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3882
      42
      SAMN12137990
      B73
      ZM4
      
      
      
      
      
      
      
      
      SRS5023572
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3883
      42
      SAMN12137989
      B73
      ZM4
      
      
      
      
      
      
      
      
      SRS5023573
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3884
      42
      SAMN12137988
      B73
      ZM3
      
      
      
      
      
      
      
      
      SRS5023570
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3885
      45
      SAMN12137987
      B73
      ZM5
      
      
      
      
      
      
      
      
      SRS5023571
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3886
      45
      SAMN12137986
      B73
      ZM5
      
      
      
      
      
      
      
      
      SRS5023568
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
    
      3887
      45
      SAMN12137985
      B73
      ZM5
      
      
      
      
      
      
      
      
      SRS5023569
      tassel primordia
      Plant sample from Zea mays subsp. mays
    
  

3888 rows × 15 columns



In [8]:

    
samples_table.to_csv('biosample_result.csv')



In [ ]:

	age	biosample_id	cultivar	dev_stage	ecotype	genotype	isolate	label	leaf_number	maize_cultivar	paragraph	source_name	sra_id	tissue	title
0	9	SAMEA5605513	B73			wild type genotype							ERS3409710		E-MTAB-7200_2:Sample 9
1	9	SAMEA5605512	B73			wild type genotype							ERS3409709		E-MTAB-7200_2:Sample 8
2	9	SAMEA5605511	B73			wild type genotype							ERS3409708		E-MTAB-7200_2:Sample 7
3	9	SAMEA5605510	B73			wild type genotype							ERS3409707		E-MTAB-7200_2:Sample 6
4	9	SAMEA5605509	B73			wild type genotype							ERS3409706		E-MTAB-7200_2:Sample 5
5	9	SAMEA5605508	B73			wild type genotype							ERS3409705		E-MTAB-7200_2:Sample 4
6	9	SAMEA5605507	B73			wild type genotype							ERS3409704		E-MTAB-7200_2:Sample 3
7	9	SAMEA5605506	B73			wild type genotype							ERS3409703		E-MTAB-7200_2:Sample 2
8	9	SAMEA5605505	B73			wild type genotype							ERS3409702		E-MTAB-7200_2:Sample 10
9	9	SAMEA5605504	B73			wild type genotype							ERS3409701		E-MTAB-7200_2:Sample 1
10		SAMEA5229560											ERS3037001		B73
11	7 days	SAMN03217742	Zea mays		B73						In this study, temporal changes of gene and mi...		SRS750388	leaf	Greening process of maize and rice leaves
12	13 days after sowing	SAMN03100283	B73										SRS718868	etiolated tissue	Zea mays subsp. mays: B73 DNA
13		SAMN03031762	B73			lg1-R							SRS700189	P6 leaf primordia	lg1-3
14		SAMN03031761	B73			lg1-R							SRS700188	P6 leaf primordia	lg1-2
15		SAMN03031760	B73			lg1-R							SRS700187	P6 leaf primordia	lg1-1
16		SAMN03031759	B73			wild type							SRS700186	P6 leaf primordia	wtL-3
17		SAMN03031758	B73			wild type							SRS700185	P6 leaf primordia	wtL-2
18		SAMN03031757	B73			wild type							SRS700184	P6 leaf primordia	wtL-1
19	2 weeks old	SAMN02996429	B73										SRS685657	shoot meristem	Plant sample from Zea mays subsp. mays
20	2 weeks old	SAMN02996428	B73										SRS685656	Shoot meristem	Plant sample from Zea mays subsp. mays
21	2 weeks old	SAMN02996427	B73										SRS685655	laser microdissected shoot meristem	Plant sample from Zea mays subsp. mays
22	2 weeks old	SAMN02996426	B73										SRS685654	Whole seedling	Plant sample from Zea mays subsp. mays
23	2 weeks	SAMN02996425	B73										SRS685653	Whole seedling	Plant sample from Zea mays subsp. mays
24		SAMN02401801											SRS502117		PARE library from Maize B73, pollen
25		SAMN02351542	inbred line Mo17					B73-stele			RNA isolated from individual tissues of young ...		SRS478255		Zea mays, Mo17 4-day-old primary roots from st...
26		SAMN02203474											SRS444558		Zea Mays, B73 x Mo17 BSA tall
27		SAMN02203473											SRS444557		Zea Mays, B73 x Mo17 BSA short
28		SAMN02203472											SRS444556		Zea Mays, B73 x Mo17 BSA late
29		SAMN02203471											SRS444555		Zea Mays, B73 x Mo17 BSA early
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
3858	36	SAMN12138014	B73	ZM2									SRS5023549	tassel primordia	Plant sample from Zea mays subsp. mays
3859	36	SAMN12138013	B73	ZM3									SRS5023547	tassel primordia	Plant sample from Zea mays subsp. mays
3860	38	SAMN12138012	B73	ZM3									SRS5023546	tassel primordia	Plant sample from Zea mays subsp. mays
3861	38	SAMN12138011	B73	ZM3									SRS5023545	tassel primordia	Plant sample from Zea mays subsp. mays
3862	38	SAMN12138010	B73	ZM4									SRS5023544	tassel primordia	Plant sample from Zea mays subsp. mays
3863	38	SAMN12138009	B73	ZM3									SRS5023543	tassel primordia	Plant sample from Zea mays subsp. mays
3864	38	SAMN12138008	B73	ZM3									SRS5023542	tassel primordia	Plant sample from Zea mays subsp. mays
3865	38	SAMN12138007	B73	ZM2									SRS5023541	tassel primordia	Plant sample from Zea mays subsp. mays
3866	39	SAMN12138006	B73	ZM3									SRS5023539	tassel primordia	Plant sample from Zea mays subsp. mays
3867	39	SAMN12138005	B73	ZM4									SRS5023540	tassel primordia	Plant sample from Zea mays subsp. mays
3868	39	SAMN12138004	B73	ZM3									SRS5023550	tassel primordia	Plant sample from Zea mays subsp. mays
3869	39	SAMN12138003	B73	ZM4									SRS5023551	tassel primordia	Plant sample from Zea mays subsp. mays
3870	40	SAMN12138002	B73	ZM4									SRS5023563	tassel primordia	Plant sample from Zea mays subsp. mays
3871	40	SAMN12138001	B73	ZM3									SRS5023565	tassel primordia	Plant sample from Zea mays subsp. mays
3872	40	SAMN12138000	B73	ZM4									SRS5023562	tassel primordia	Plant sample from Zea mays subsp. mays
3873	41	SAMN12137999	B73	ZM5									SRS5023564	tassel primordia	Plant sample from Zea mays subsp. mays
3874	41	SAMN12137998	B73	ZM5									SRS5023560	tassel primordia	Plant sample from Zea mays subsp. mays
3875	41	SAMN12137997	B73	ZM4									SRS5023561	tassel primordia	Plant sample from Zea mays subsp. mays
3876	41	SAMN12137996	B73	ZM4									SRS5023558	tassel primordia	Plant sample from Zea mays subsp. mays
3877	42	SAMN12137995	B73	ZM5									SRS5023559	tassel primordia	Plant sample from Zea mays subsp. mays
3878	42	SAMN12137994	B73	ZM5									SRS5023576	tassel primordia	Plant sample from Zea mays subsp. mays
3879	42	SAMN12137993	B73	ZM5									SRS5023577	tassel primordia	Plant sample from Zea mays subsp. mays
3880	42	SAMN12137992	B73	ZM5									SRS5023574	tassel primordia	Plant sample from Zea mays subsp. mays
3881	42	SAMN12137991	B73	ZM4									SRS5023575	tassel primordia	Plant sample from Zea mays subsp. mays
3882	42	SAMN12137990	B73	ZM4									SRS5023572	tassel primordia	Plant sample from Zea mays subsp. mays
3883	42	SAMN12137989	B73	ZM4									SRS5023573	tassel primordia	Plant sample from Zea mays subsp. mays
3884	42	SAMN12137988	B73	ZM3									SRS5023570	tassel primordia	Plant sample from Zea mays subsp. mays
3885	45	SAMN12137987	B73	ZM5									SRS5023571	tassel primordia	Plant sample from Zea mays subsp. mays
3886	45	SAMN12137986	B73	ZM5									SRS5023568	tassel primordia	Plant sample from Zea mays subsp. mays
3887	45	SAMN12137985	B73	ZM5									SRS5023569	tassel primordia	Plant sample from Zea mays subsp. mays