Handling the input matrix in CB2

In CB2: CRISPR Pooled Screen Analysis using Beta-Binomial Test

If an analysis starts with an input matrix, it has to be appropriately pre-proceed before it is used any functions of CB2. CB2 allows two different types of input: a numeric matrix/data frame with row.names and a data.frame contains columns of counts and columns of sgRNA IDs and target genes. Either of them will work. This document explains how the input should be formed and how to process the input using CB2. In the entire document, [@evers2016crispr]'s CRISPR-RT112 screen data are used.

knitr::opts_chunk$set(echo = TRUE)

The following code imports required packages which are required to run below codes.

library(CB2)
library(dplyr)
library(readr)

The following code block shows an example of the first type of input which CB2 can handle. Each column of Evers_CRISPRn_RT112$count contains counts of guide RNAs of a sample (that was initially extracted from NGS data). A count of the input shows that how many guide RNA barcodes were observed from a given NGS sample. Each row of the matrix has a row name (e.g., RPS19_sg10), and the name is the ID of a guide RNA. For example, RPS19_sg10, which is the first-row name in the example, indicates that the first row contains the counts of RPS_sg10 guide RNA. Every guide RNA ID must have exactly one _ character, and it is used to be a separator of two strings. The first string displays the name of a gene whose gene is targeted by the guide RNA, and the second string is used as an identifier among guide RNAs that targets the same gene. For example, RPS_sg10 indicates that the guide RNA is designed to target the RPS gene, and sg10 is the unique identifier.

NOTE : If the input contains multiple _ characters, CB2 is not able to run. In particular, if Entrez gene IDs are used as the gene names, CB2 does not handle the input. One of the solutions for this case is changing the gene names to another identifier (e.g., HGNC symbol) or using another type of input, which will explain below.

data("Evers_CRISPRn_RT112")
head(Evers_CRISPRn_RT112$count)

In addition, CB2 requires experiment design information which is formed as a data.frame and contains sample names and groups of each sample. In Evers_CRISPRn_RT112 data, Evers_CRISPRn_RT112$design is the data.frame.

Evers_CRISPRn_RT112$design

With the two variables, CB2 can perform the hypothesis test with measure_sgrna_stats and measure_gene_stats functions.

sgrna_stats <- measure_sgrna_stats(Evers_CRISPRn_RT112$count, Evers_CRISPRn_RT112$design, "before", "after")
gene_stats <- measure_gene_stats(sgrna_stats)
head(gene_stats)

Another input type is a data.frame that contains two additional columns, which contain the guide RNA information (target gene and guide RNA identifier). A CSV file which was used in the CB2 publication ([@jeong2019beta]). The CSV file contains the additional columns, the first is the gene column, and the second is the sgRNA column.

df <- read_csv("https://raw.githubusercontent.com/hyunhwaj/CB2-Experiments/master/01_gene-level-analysis/data/Evers/CRISPRn-RT112.csv")
df

Two additional parameters have to be set to the measure_sgrna_stats function if an input matrix is this type. The first parameter is ge_id, which specifies the column of genes, and the second parameter is sg_id, which specifies the column of IDs. In the following code, gene_id sets as gene and sg_id sets as sgRNA.

head(measure_sgrna_stats(df, Evers_CRISPRn_RT112$design, "before", "after", ge_id = 'gene', sg_id = 'sgRNA'))

References

CB2 documentation built on July 24, 2020, 5:08 p.m.