I would like to convert the fastq files from this dataset: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE98679 to .bam files. The group did not provide the scripts they used to to their alignment, I know they already posted their reads but I need the .bam files for my pipeline. Is it possible to do this without knowing the cell barcodes? If so with what tools?
The linked dataset is based on the Smart-seq2 technology. This is a plate-based assay in which individual cells are first sorted into individual wells of a microwell plate via FACS. The cells are then individually lysed and libraries are prepared. Since this happens per well one assigns a single sequencing barcode to each well/cell, hence no further cellular barcoding as in droplet-based methods such as 10x Chromatium is required. Long story short, each fastq file that one obtains from the sequencing represents one individual cell. Hence, one can use any standard RNA-seq alignment or quantification technique such as
salmon to obtain alignments/quantifications. One would then need to parse the individual quantifications into a single count matrix. If one used aligners such as
STAR this could be done with high-performance quantifiers such as
featureCounts. If using pseudo- or selective aligners such as
kallisto I would use
tximport to aggregate the per-transcript quantifications to the gene level while importing into R to get a gene-level count matrix. As each file is a single cell there is no need for more complicated solutions such as
CellRanger which do not apply here.