Dataset processing

Above: Rule graph for complete data processing

Obtaining raw data

Raw data can come from two supported sources:

• From parallel-fastq-dump
• From FTP (fastq files and bams)

The rule is determined based on the determined meta-information obtained from get_all_meta (see how we obtain meta information).

Rules to obtain raw data are: get_fastq_dump_files for parallel-fastq-dump, and get_fastq_ftp, get_bam_ftp for FTP files.

STARSolo

We then generate required STAR index (get_star_index) using FASTA and GTF files from cellranger index (if index is not generated yet) for desired species and generate bash script (render_star_script) which will combine all raw files from runs that correspond to single sample.

To see which parameters are used when we generate star_script, please, refer to worfklow/templates/star.bash which is Jinja2 template we use to render the script.

We then run the generated script (run_star) to obtain gene/cell expression matrix for downstream analysis.

We are currently using unmodified filtered output matrix from STARSolo as an input for seurat, but we might change this in future in filter_counts_star if we decide that there are better approaches to filter empty droplets from the matrix.