genedataset is a package to store and access gene expression datasets and gene definitions. It consists of two main classes, geneset and dataset.
Some significant changes have been made in this version:
- "MedianTranscriptLength" property of Geneset has been replaced with "TranscriptLengths", which holds a list of transcript lengths, one per transcript (eg. [2000,1530]). Note that the value contained is a string, so convert into a list of integers before using the value.
- The Gene annotation has been upgraded to Ensembl version 88.
- Dataset no longer supports microarrays, so it's been simplified, so related methods are deprecated.
- The package supports both python 3 and 2 - tested on 2.7.14 and 3.6.3.
pip install genedataset
# If this installs the old version 0.6 instead of the latest, explicitly specify the version:
pip install genedataset==[latest_version_number]geneset stores gene information combined from both Ensembl and NCBI/Entrez (mouse and human only), so that you can query it:
from genedataset import geneset
gs = geneset.Geneset().subset(queryStrings='ccr3')
print(gs.geneIds())
['ENSG00000183625', 'ENSMUSG00000035448']
gs.dataframe()| EnsemblId | Species | EntrezId | GeneSymbol | Synonyms | Description | TranscriptLengths | Orthologue |
|---|---|---|---|---|---|---|---|
| ENSG00000183625 | HomoSapiens | 1232 | CCR3 | CC-CKR-3 | CD193 | CKR3 | CMKBR3 |
| ENSMUSG00000035448 | MusMusculus | 12771 | Ccr3 | CC-CKR3 | CKR3 | Cmkbr1l2 | Cmkbr3 |
dataset can store gene expression data so that it can be queried. The stored data consists of expression values (usually rna-seq) and sample data packaged into HDF5 format.
from genedataset import dataset
ds = dataset.Dataset("genedataset/data/testdata.1.0.h5")
ds
<Dataset name:testdata 4 samples>
ds.expressionMatrix()| featureId | s01 | s02 | s03 | s04 |
|---|---|---|---|---|
| gene1 | 3.45 | 4.65 | 2.65 | 8.23 |
| gene2 | 5.54 | 0.00 | 1.43 | 6.43 |
| gene3 | 0.00 | 0.00 | 4.34 | 5.44 |
ds.sampleTable()| sampleId | celltype | tissue |
|---|---|---|
| s01 | B1 | BM |
| s02 | B1 | BM |
| s03 | B2 | BM |
| s04 | B2 | BM |
Here is an example to create a Dataset file from text files. Once the file has been created, it can be accessed through the Dataset instance. The advantage of this is to store all related information for a dataset in one file, and gives you a python object that can be used for analyses and for application development.
import pandas
from genedataset import dataset
attributes = {"name": "testdata",
"fullname": "Test Dataset",
"version": "1.0",
"description": "This dataset comes with the package for testing purposes.",
"expression_data_keys": ["counts","cpm"],
"pubmed_id": None,
"species": "MusMusculus"}
samples = pandas.DataFrame([['B1', 'BM'], ['B1', 'BM'], ['B2', 'BM'], ['B2', 'BM']],
index=['s01','s02','s03','s04'],
columns=['celltype','tissue'])
samples.index.name = "sampleId"
counts = pandas.DataFrame([[35, 44, 21, 101], [50, 0, 14, 62], [0, 0, 39, 73]],
index=['gene1', 'gene2', 'gene3'], columns=['s01', 's02', 's03', 's04'])
counts.index.name = "geneId"
cpm = pandas.DataFrame([[3.45, 4.65, 2.65, 8.23], [5.54, 0.0, 1.43, 6.43], [0.0, 0.0, 4.34, 5.44]],
index=['gene1', 'gene2', 'gene3'], columns=['s01', 's02', 's03', 's04'])
cpm.index.name = "geneId"
dataset.createDatasetFile("/datasets", attributes=attributes, samples=samples, expressions=[counts,cpm])- e-mail: jarnyc@unimelb.edu.au
- v1.0 - Major upgrade.
- v0.1.x - Initial release with minor adjustments to test pypi and github upload/download.