First, for gene G, I want to create a pandas dataframe for control and experimental conditions, where the ratio of 0:1 is 10% and 20%, respectively.

import pandas as pd

n = 5000

df = pd.DataFrame.from_dict(
     {"Cells": (f'Cell{x}' for x in range(1, n 1)), "Control": np.random.choice([1,0], p=[0.1, 0.9], size=n), "Experimental": np.random.choice([1,0], p=[0.1 0.1, 0.9-0.1], size=n)},
     orient='columns'
)

df = df.set_index("Cells")

Second, I perform cross-tabulation analysis for gene G.

# Contingency table/array
table = pd.crosstab(df.index, [df["Control"], df["Experimental"]])
table

Now, I want to extrapolate the conditions from step 1 and 2 for 1000 genes and then perform cross-tabulation. How?

CodePudding user response：

There are multiple ways to handle it. One of the easiest would be:

1. Wrap the simulation procedure in a function and run it for each gene to create a list of genes based results.
2. Concat all the gene_based simulations and run the crosstab function to get the contingency tables in a single dataframe.

Something like:

n = 100   # no of records per simulation/ gene
nog = 2   # no of genes
gene_list = ["gene_"   str(i) for i in range(0,nog)]

# simulation procedure
def generate_gene_df(gene, n):

    df = pd.DataFrame.from_dict(
         {"Gene" : gene,
          "Cells": (f'Cell{x}' for x in range(1, n 1)),
          "Control": np.random.choice([1,0], p=[0.1, 0.9], size=n),
          "Experimental": np.random.choice([1,0], p=[0.1 0.1, 0.9-0.1], size=n)},
         orient='columns'
    )

    df = df.set_index(["Gene","Cells"])
    
    return df

# create a list of gene based simulations generated using your procedure
gene_df_list = [generate_gene_df(gene, n) for gene in gene_list]

single_genes_df = pd.concat(gene_df_list)
single_genes_df = single_genes_df.reset_index()


table = pd.crosstab([single_genes_df["Gene"], single_genes_df["Cells"]], 
                    [single_genes_df["Control"], single_genes_df["Experimental"]])
table