Tutorial 1: CPTAC Data Introduction¶

The National Cancer Institute’s Clinical Proteomic Tumor Analysis Consortium (CPTAC) is a national effort to accelerate the understanding of the molecular basis of cancer through the application of large-scale proteome and genome analysis, or proteogenomics. CPTAC generates comprehensive proteomics and genomics data from clinical cohorts, typically with ~100 samples per tumor type. The graphic below summarizes the structure of each CPTAC dataset. For more information, visit the NIH website.

This Python package makes accessing CPTAC data easy with Python code and Jupyter notebooks. The package contains several tutorials which demonstrate data access and usage. This first tutorial serves as an introduction to the data to help users become familiar with what is included and how it is presented.

Data Overview¶

Our package provides data access in a Python programming environment. If you have not installed Python or have not installed the package, see our installation documentation here.

Once we have the package installed and we're in our Python environment, we begin by importing the package with a standard Python import statement:

To view the available datasets, call the cptac.list_datasets() function:

Data Availability¶

The goals of CPTAC as a consortium include the broad and open dissemination of cancer proteogenomic data. The timing of the a dataset's public release generally follows three stages: internal release to CPTAC investigators, public release with a publication embargo, and full public release. Each of the cancer types may be at a different data availability stage, depending on the date of data creation. In the Python cptac package, these three stages are dealt with as follows:

Internally released data requires a password to download.

Embargoed release data is publicly available, but prints an embargo statement every time you interact with the data.

Public data is fully released without restrictions.

Downloading data¶

The cptac package stores the data files for each dataset on a remote server. When you first install cptac, you will have no data files. To install the latest version of the data files for a particular dataset, simply call the cptac.download function, passing the name of your desired dataset for the dataset parameter:

Exploring the data¶

Once you've downloaded a dataset, cptac allows you to load the dataset into a Python variable, and you can use that variable to access and work with the data. To load a particular dataset into a variable, type the name you want to give the variable, followed by =, and then type cptac. and the name of the dataset in UpperCamelCase followed by two parentheses, e.g. cptac.Endometrial() or cptac.Ccrcc():

To see what data is available, use the en.list_data() function. This displays the different types of data included in the dataset for this particular cancer type, each stored in a pandas dataframe. It also prints the dimensions of each dataframe.

Molecular Omics¶

Data can be accessed through several "get" functions. For example, we can look at the proteomics data by using en.get_proteomics(). This returns a pandas dataframe containing the proteomic data. Each column in the proteomics dataframe is the quantitiative measurement for a particular protein. Each row in the proteomics dataframe is a sample of either a tumor or non-tumor from a cancer patient.

Dataframe values¶

Values in the dataframe are protein abundance values. Values that read "NaN" mean that particular sample from that patient had no data for that particular protein. For the endometrial CPTAC proteomics data, a TMT-reference channel strategy was used. A detailed description of this strategy can be found at Nature Protocols and also at PubMed Central. This strategy ratios each sample's abundance to a pooled reference. The ratio is then log transformed. Therefore positive values indicate a measurement higher than the pooled reference; negative values are lower than the pooled reference.

As seen in en.list_data(), other omics data are also available (e.g. transcriptomics, copy number variation, phoshoproteomics).

The transcriptomics looks almost identical to the proteomics data, available in a pandas dataframe with the same convention. Each set of samples is consitent, meaning samples found in the endometrial proteomics data will be the same samples in all other endometrial dataframes.

Clinical Data¶

The clinical dataframe lists clinical information for the patient associated with each sample (e.g. age, race, diabetes status, tumor size).

In addition to donating a tumor sample, some patients also had a normal sample taken for control and comparison. We can identify these samples by looking for samples marked "Normal" in the "Sample_Tumor_Normal" column, and whose Patient IDs are the same as the Patient IDs of tumor samples, but with a ".N" appended to the ID. For example, patient C3L-00006 provided both a tumor sample (marked C3L-00006) and a normal sample (marked C3L-00006.N). Note that the normal samples do not have many values in the clinical columns, because much of the information does not apply to non-tumor samples. Additionally, in cases where a column would have identical values for tumor and normal samples from the same patient (e.g., patient age and gender), the information is recorded only for the tumor sample.

Mutation data¶

Each cancer dataset contains mutation data for the cohort. The data consists of all somatic mutations found for each sample (meaning there will be many lines for each sample). Each row lists the specific gene that was mutated, the type of mutation, and the location of the mutation. This data is a direct import of a MAF file.

Exporting dataframes¶

If you wish to export a dataframe to a file, simply call the dataframe's to_csv method, passing the path you wish to save the file to, and the value separator you want:

Getting help with a dataset or function¶

To view the documentation for a dataset, pass it to the Python help function, e.g. help(en). You can also view the documentation for just a specific function: help(en.join_omics_to_omics).