Investigate artifacts produced by cell dissociation methods that confound scRNA-seq analysis
Set up workspace by running ./setup-dir.sh which will create the following nested structure:
scripts/
- scripts/01_process-bulkrna
- scripts/02_process-scrna
- scripts/03_compare-bulk_scrna
data/
- data/01_process-bulkrna
- data/02_process-scrna
- data/03_compare-bulk_scrna
raw-data/
sandbox/
notes/
Pipeline largely relies on docker and cwltools to be installed. Make sure to have these installed prior to running this pipeline. Some examples include pip install cwltool and rather than sudo apt install cwltool in case you have multiple CWL implementations.
Building a reference genome requires reference Ensembl files and this pipeline assumes you have already retrieved these. The existing pipeline uses the following files (but if you would like to use different files then be sure to update the scripts directly - as these files have been hardcoded)
-
Fasta for STAR
Homo_sapiens.GRCh38.dna.primary_assembly.fa.gz -
GTF for STAR and featureCounts
wget ftp://ftp.ensembl.org/pub/release-99/gtf/homo_sapiens/Homo_sapiens.GRCh38.98.gtf.gz
PURPOSE: explanation of raw sequencing files location
Click for details
Medgenome ftp. Files were moved into raw-data dir using wget -r --no-parent --password <pwd> https://portal.us.medgenome.com/<dir>/ --no-check-certificate --user <user>
Last updated 2/26/20
Samples paired with P2000996_02292020
2 patient samples. Recieved seq files on Feb. 25, 2020. This does NOT include the bulk RNA-seq data. That data will be in a separate dir.
Fq files:
7319-Enz_S5_L002_I1.fastq.gz
7319-Enz_S5_L002_R1.fastq.gz
7319-Enz_S5_L002_R2.fastq.gz
7319-Mech_S4_L002_I1.fastq.gz
7319-Mech_S4_L002_R1.fastq.gz
7319-Mech_S4_L002_R2.fastq.gz
7320E_S7_L002_I1.fastq.gz
7320E_S7_L002_R1.fastq.gz
7320E_S7_L002_R2.fastq.gz
7320M_S6_L002_I1.fastq.gz
7320M_S6_L002_R1.fastq.gz
7320M_S6_L002_R2.fastq.gz
Last updated 3/2/20
Samples that are paired with P2000997_02242020. Received on March 2, 2020.
Fq files:
# Samples bulk rna
SCC_7319_B2_R1.fastq.gz #patient 7319
SCC_7319_B2_R2.fastq.gz #patient 7319
SCC_7320_B2_R1.fastq.gz #patient 7320
SCC_7320_B2_R2.fastq.gz #patient 7320
# Samples ground truth
SCC_7319_GT_R1.fastq.gz #patient 7319
SCC_7319_GT_R2.fastq.gz #patient 7319
SCC_7320_GT_R1.fastq.gz #patient 7319
SCC_7320_GT_R2.fastq.gz #patient 7319
# N fq
SCC_7320_N_R1.fastq.gz
SCC_7320_N_R2.fastq.gz
Last updated 2/26/20
All additional data outside of our project. Will be used for deconvolution models
PURPOSE: to take raw sequencing reads and convert to a gene count matrix
Base script dir scripts/01_process-bulkrna/
Click for details
The dir scripts/01_process-bulkrna/ contains the scripts for processing raw sequence reads of bulk RNA-seq samples through generating count matrices and inspecting count matrices for odd occurances
The wrapper script allows for batch scripting tracking wrapper.sh
Step 1:
- Genome indexing (create alignment indices)
Step 2:
- Demultiplex reads - skip because seq core did this for us
- Read quality control
- Adapter and poor quality read trimming
- Check read quality
- Alignment
- Alignment File QC
- Generate count matrices - from aligned reads (featureCounts) -> called
${basename}_COUNTmatrix.txt
STAR aligner. Job ID 12460222
Job IDs:
- SCC_7319_B2_R1.fastq.gz – job id 12515579
- SCC_7319_B2_R2.fastq.gz – job id 12515580
- SCC_7319_GT_R1.fastq.gz – job id 12515709
- SCC_7320_GT_R1.fastq.gz – job id 12515710
Sequencing core completed this for us. Therefore not included in our built pipeline.
Produce read quality reports of raw seq files (that have been demultiplexed).
Requires manual inspection of two output summary files to determine input parameters for read trimming.
[TODO] Adapter sequence trimming will be added once have seq results from the Core
Trimmomatic 0.39. Paired trimming. Although virtually all adapter sequences should already have been trimmed, we will conduct a secondary pass to remove any remaining adapter sequences. Then low quality reads will be trimmed.
Produce read quality reports of trimmed seq files.
STAR aligner
Each FASTQ file has 5 output files, including unsorted by name BAM file
[TODO] Adapter sequence trimming will be added once have seq results from the Core
Samtools
featureCounts (gene-level counting) and produces final count matrix ${basename}_COUNTmatrix.txt
- Parameters set for stranded, ignore multi-mapping reads, not current for paired end data
Note Transcript-level quantification is less accurate than gene-level quantification (e.g. salmon, RSEM, kallisto). Also transcript-level quantification has less clear biological interpretability. Less statistical power if split counts between isoforms.
Example CWL included in this scripts dir and will be implemented for publication. For now it is simply placed there as an example.
PURPOSE: to take raw sequencing reads and convert to a gene-cell count matrix
Base script dir scripts/02_process-scrna/
Click for details
Scripts here
PURPOSE: to identify cell populations through gene marker analysis in single cell data. In otherwords, comparing profile of mechanical dissociation scRNA-seq to enzymatic scRNA-seq
Base script dir scripts/03_compare-bulk_scrna/signatures/
Click for details
Scripts here
PURPOSE: to convert scRNA gene-cell count matrices into a pseudo-bulk count matrix. Then compare this pseduo-bulk data to the bulk RNA count matrix
Base script dir scripts/03_compare-bulk_scrna/scrna2bulk/
Click for details
Analysis hardcoded in dge.Rmd and the rendered version dge.html
PURPOSE: to convert bulk RNA count matrices into a pseudo-scRNA count matrix - this will be achieved by the cell-wise predictions of several deconvolution algorithms. Then compare this pseduo-scRNA data to the scRNA gene-cell count matrix
Base script dir scripts/03_compare-bulk_scrna/deconv/
Click for details
Scripts here