processAmplicons: Process raw data from pooled genetic sequencing screens

Description

Given a list of sample-specific index (barcode) sequences and hairpin/sgRNA-specific sequences from an amplicon sequencing screen, generate a DGEList of counts from the raw fastq file/(s) containing the sequence reads. Assumes fixed structure of amplicon sequences (i.e. both the sample-specific index sequences and hairpin/sgRNA sequences can be found at particular locations within each read).

Usage

processAmplicons(readfile, readfile2=NULL, barcodefile, hairpinfile,
                    barcodeStart=1, barcodeEnd=5, 
                    barcode2Start=NULL, barcode2End=NULL,
                    barcodeStartRev=NULL, barcodeEndRev=NULL, 
                    hairpinStart=37, hairpinEnd=57,
                    allowShifting=FALSE, shiftingBase=3,
                    allowMismatch=FALSE, barcodeMismatchBase=1, 
                    hairpinMismatchBase=2, allowShiftedMismatch=FALSE, 
                    verbose=FALSE)

Arguments

readfile

character vector giving one or more fastq filenames

readfile2

character vector giving one or more fastq filenames for reverse read, default to NULL

barcodefile

filename containing sample-specific barcode ids and sequences

hairpinfile

filename containing hairpin/sgRNA-specific ids and sequences

barcodeStart

numeric value, starting position (inclusive) of barcode sequence in reads

barcodeEnd

numeric value, ending position (inclusive) of barcode sequence in reads

barcode2Start

numeric value, starting position (inclusive) of second barcode sequence in forward reads

barcode2End

numeric value, ending position (inclusive) of second barcode sequence in forward reads

barcodeStartRev

numeric value, starting position (inclusive) of barcode sequence in reverse reads, default to NULL

barcodeEndRev

numeric value, ending position (inclusive) of barcode sequence in reverse reads, default to NULL

hairpinStart

numeric value, starting position (inclusive) of hairpin/sgRNA sequence in reads

hairpinEnd

numeric value, ending position (inclusive) of hairpin/sgRNA sequence in reads

allowShifting

logical, indicates whether a given hairpin/sgRNA can be matched to a neighbouring position

shiftingBase

numeric value of maximum number of shifted bases from input hairpinStart and hairpinEnd should the program check for a hairpin/sgRNA match when allowShifting is TRUE

allowMismatch

logical, indicates whether sequence mismatch is allowed

barcodeMismatchBase

numeric value of maximum number of base sequence mismatches allowed in a barcode sequence when allowShifting is TRUE

hairpinMismatchBase

numeric value of maximum number of base sequence mismatches allowed in a hairpin/sgRNA sequence when allowShifting is TRUE

allowShiftedMismatch

logical, effective when allowShifting and allowMismatch are both TRUE. It indicates whether we check for sequence mismatches at a shifted position.

verbose

if TRUE, output program progress

Value

Returns a DGEList object with following components:
countsread count matrix tallying up the number of reads with particular barcode and hairpin/sgRNA matches. Each row is a hairpin and each column is a sample
genesIn this case, hairpin/sgRNA-specific information (ID, sequences, corresponding target gene) may be recorded in this data.frame
lib.sizeauto-calculated column sum of the counts matrix

Details

The processAmplicons function assumes the sequences in your fastq files have a fixed structure (as per Figure 1A of http://f1000research.com/articles/3-95/v2). It cannot be used if your hairpins/sgRNAs/sample index sequences are in random locations within each read. You will need to customise your own sequence processing pipeline if this is the case, but can still complete your downstream analysis using edgeR. The input barcode file and hairpin/sgRNA files are tab-separated text files with at least two columns (named 'ID' and 'Sequences') containing the sample or hairpin/sgRNA ids and a second column indicating the sample index or hairpin/sgRNA sequences to be matched. If barcode2Start and barcode2End are specified, a third column 'Sequences2' is expected in the barcode file. If readfile2, barcodeStartRev and barcodeEndRev are specified, another column 'SequencesReverse' is expected in the barcode file. The barcode file may also contain a 'group' column that indicates which experimental group a sample belongs to. Additional columns in each file will be included in the respective $samples or $genes data.frames of the final code{DGEList} object. These files, along with the fastq file/(s) are assumed to be in the current working directory. To compute the count matrix, matching to the given barcodes and hairpins/sgRNAs is conducted in two rounds. The first round looks for an exact sequence match for the given barcode sequences and hairpin/sgRNA sequences at the locations specified. If allowShifting is set to TRUE, the program also checks if a given hairpin/sgRNA sequence can be found at a neighbouring position in the read. If a match isn't found, the program performs a second round of matching which allows for sequence mismatches if allowMismatch is set to TRUE. The program also checks parameter allowShiftedMismatch which accommodates mismatches at the shifted positions. The maximum number of mismatch bases in barcode and hairpin/sgRNA are specified by the parameters barcodeMismatchBase and hairpinMismatchBase. The program outputs a DGEList object, with a count matrix indicating the number of times each barcode and hairpin/sgRNA combination could be matched in reads from input fastq file/(s). For further examples and data, refer to the Case studies available from http://bioinf.wehi.edu.au/shRNAseq/. Note: The processAmplicons function supercedes the earlier processHairpinReads function.

References

Dai Z, Sheridan JM, et al. (2014). edgeR: a versatile tool for the analysis of shRNA-seq and CRISPR-Cas9 genetic screens. F1000Research, 3:95. http://f1000research.com/articles/3-95/v2. PMID: 24860646.