Use Case 1: Comparing Omics Data

Step 1: Importing packages and setting up your notebook.

This use case will guide you through basic features of our package, including importing the package and loading the data, interacting with the dataframes, and producing a basic scatterplot comparing proteomics and transcriptomics for a given gene. We will begin by importing standard data analysis libraries.

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

Next, we will import the cptac package, and load the Endometrial dataset object. This dataset object includes functions that allow the user to access and perform basic operations on the data from the endometrial cancer patients.

import cptac
import cptac.utils as ut

en = cptac.Ucec()

Step 2: Joining dataframes

Now that we've imported the package and loaded the data, we can begin to extract the data we need. In this use case, we want to compare the proteomics and transcriptomics data from the endometrial cancer patients for the A1BG gene. For a full list of available data, enter en.list_data_sources(). For a full list of functions available for accessing and manipulating the data, enter help(en). To access documentation for a specific function, pass it to the help() function, e.g. help(en.multi_join).

We use the en.multi_join() function to join the A1BG columns from the proteomics and transcriptomics dataframes into a single dataframe. Note when we are using multi-join that you specify the datasets as a dictionary. Keys are the source and datatype, values are the gene(s) of interest.

A1BG_cross = en.multi_join({"umich proteomics": "A1BG", "bcm transcriptomics": "A1BG"})

# Use cptac.utils.reduce_multiindex to get rid of a level we don't need from the column index
A1BG_cross = ut.reduce_multiindex(A1BG_cross, levels_to_drop="Database_ID")

print(A1BG_cross.head())

cptac warning: Your version of cptac (1.5.1) is out-of-date. Latest is 1.5.0. Please run 'pip install --upgrade cptac' to update it. (C:\Users\sabme\anaconda3\lib\threading.py, line 910)

Name        A1BG_umich_proteomics  A1BG_bcm_transcriptomics
Patient_ID                                                 
C3L-00006               -1.121101                      2.54
C3L-00008               -0.798504                      4.40
C3L-00032               -0.577203                      4.83
C3L-00084                1.612713                      4.73
C3L-00090               -1.350755                      4.14

Step 3: Plot data

We then plot this data to visually explore the correlation between proteomics and transcriptomics data. For the A1BG gene, we see a positive correlation, indicating that higher transcript levels correlate with higher protein levels in the patients.

sns.set(style="darkgrid")
plot = sns.regplot(x=A1BG_cross.columns[0], y=A1BG_cross.columns[1], 
                   data=A1BG_cross)
plot.set(xlabel='Proteomics', ylabel='Transcriptomics', 
         title='Proteomics vs. Transcriptomics for the A1BG gene')
plt.show()

Step 4: Plot more data

We apply the same analysis to the RPL11 gene, another gene of interest in cancer research. The plot visualizes any correlation between mRNA and protein abundance for this gene in the data.

gene = 'RPL11'
gene_cross = en.multi_join({"umich proteomics": gene, "bcm transcriptomics": gene});
# Use cptac.utils.reduce_multiindex to get rid of a level we don't need from the column index
gene_cross = ut.reduce_multiindex(gene_cross, levels_to_drop="Database_ID")
plot = sns.regplot(x=gene_cross.columns[0], y=gene_cross.columns[1], data=gene_cross, color="green")
plot.set(xlabel='Proteomics', ylabel='Transcriptomics',
         title='Proteomics vs. Transcriptomics for the ' + gene + ' gene')
plt.show()

Step 5: Repeat with the Ovarian Dataset

The same analysis can be applied to different datasets, such as ovarian cancer data. We perform the analysis on the PTEN gene, a tumor suppressor gene often mutated in many types of cancer.

ov = cptac.Ov()
ovarian_cross = ov.multi_join({"umich proteomics": "PTEN", "bcm transcriptomics": "PTEN"})

# Use cptac.utils.reduce_multiindex to get rid of a level we don't need from the column index
ovarian_cross = ut.reduce_multiindex(ovarian_cross, levels_to_drop="Database_ID")

sns.set(style="darkgrid")
plot = sns.regplot(x=ovarian_cross.columns[0], y=ovarian_cross.columns[1], data=ovarian_cross)
plot.set(xlabel='Proteomics', ylabel='Transcriptomics', title='Proteomics vs. Transcriptomics for the PTEN gene')
plt.show()

Step 6: Repeat with the Colon dataset

We demonstrate the versatility of this analysis approach by applying it to colon cancer data and focusing on the SOX9 gene, which plays a role in cell differentiation and is implicated in many cancers.

co = cptac.Coad()
source1 = 'umich'
source2 = 'bcm'
colon_cross = co.join_omics_to_omics(df1_name="proteomics", df2_name="transcriptomics",
                                     df1_source=source1, df2_source=source2,
                                     genes1="SOX9", genes2="SOX9")

sns.set(style="darkgrid")
plot = sns.regplot(x=colon_cross.columns[0], y=colon_cross.columns[1], data=colon_cross)
plot.set(xlabel='Proteomics', ylabel='Transcriptomics', title='Proteomics vs. Transcriptomics for the SOX9 gene')
plt.show()