orange-bioinformatics / docs / rst / reference / geo.rst

Full commit

NCBI's Gene Expression Omnibus interface (:mod:`obiGEO`)

:obj:`obiGEO` provides an interface to NCBI's Gene Expression Omnibus repository. Currently, it only supports GEO DataSets information querying and retrieval.

The following illustrates how :obj:`GDS.getdata` is used to construct a data set with genes in rows and samples in columns. Notice that the annotation about each sample is retained in .attributes.

>>> from import obiGEO
>>> gds = obiGEO.GDS("GDS1676")
>>> data = gds.getdata()
>>> len(data)
>>> data[0]
[?, ?, -0.803, 0.128, 0.110, -2.000, -1.000, -0.358], {"gene":'EXO1'}
>>> data.domain.attributes[0]
FloatVariable 'GSM63816'
>>> data.domain.attributes[0].attributes
Out[191]: {'dose': '20 U/ml IL-2', 'infection': 'acute ', 'time': '1 d'}

GDS classes

An example that uses obj:GDSInfo:

>>> from Orage import obiGEO
>>> info = obiGEO.GDSInfo()
>>> info.keys()[:5]
>>> ['GDS2526', 'GDS2524', 'GDS2525', 'GDS2522', 'GDS1618']
>>> info['GDS2526']['title']
'c-MYC depletion effect on carcinoma cell lines'
>>> info['GDS2526']['platform_organism']
'Homo sapiens'


The following script prints out some information about a specific data set. It does not download the data set, just uses the (local) GEO data sets information file (:download:` <code/>`).

The output of this script is:

Features: 39114
Genes: 19883
Organism: Mus musculus
PubMed ID: 11827943
Sample types:
  disease state (diabetic, diabetic-resistant, nondiabetic)
  strain (NOD, Idd3, Idd5, Idd3+Idd5, Idd9, B10.H2g7, B10.H2g7 Idd3)
  tissue (spleen, thymus)

Examination of spleen and thymus of type 1 diabetes nonobese diabetic
(NOD) mouse, four NOD-derived diabetes-resistant congenic strains and
two nondiabetic control strains.

GEO data sets provide a sort of mini ontology for sample labeling. Samples belong to sample subsets, which in turn belong to specific types. Like above GDS10, which has three sample types, of which the subsets for the tissue type are spleen and thymus. For supervised data mining it would be useful to find out which data sets provide enough samples for each label. It is (semantically) convenient to perform classification within sample subsets of the same type. We therefore need a script that goes through the entire set of data sets and finds those, where there are enough samples within each of the subsets for a specific type. The following script does the work. The function valid determines which subset types (if any) satisfy our criteria. The number of requested samples in the subset is by default set to n=40 (:download:` <code/>`).

The requested number of samples, n=40, seems to be a quite a stringent criteria met - at the time of writing of this documentation - by 35 sample subsets. The output starts with:

  genotype/variation: wild type/48, upf1 null mutant/48
  cell type: macrophage/48, monocyte/48
  protocol: training set/46, validation set/47
  protocol: PUFA consumption/42, SFA consumption/42
  agent: vehicle, control/46, TCE/48
  other: non-neural/50, neural/100
  genotype/variation: wild type/56, Nrf2 null/54
  agent: untreated/55, insulin/55

Let us now pick data set GDS2960 and see if we can predict the disease state. We will use logistic regression, and within 10-fold cross validation measure AUC, the area under ROC. AUC is the probably for correctly distinguishing between two classes if picking the sample from target (e.g., the disease) and non-target class (e.g., control). From (:download:` <code/>`)

The output of this script is:

Samples: 101, Genes: 4068
AUC = 0.960

The AUC for this data set is very high, indicating that using these gene expression data it is almost trivial to separate the two classes.