Question 1

Which assay types does the Quality Control pipeline support?

Accepted Answer

FastQC runs on Read 1 and Read 2 of every sample, so any assay that arrives as raw FASTQ gets a documented quality check. Three checks are targeted: 10x chemistry detection and the preliminary single-cell QC metrics run on single-cell samples from human, mouse, rat, monkey or zebrafish, and the strandedness call runs on human bulk RNA-seq samples. Other assay types receive their FastQC results and the resource recommendations table.

Question 2

What do I get back from a Quality Control run?

Accepted Answer

Four things: FastQC results for every sample and read, a resource recommendations table with CPU and memory settings for the pipeline that follows, single-cell QC and library-type tables where those checks applied, and one MultiQC report per experiment with the recommendation and QC tables at the top.

Question 3

How do I read the MultiQC report?

Accepted Answer

Start at the top. The recommendation and QC tables line every sample up side by side, so the first pass is a scan for the sample that looks different from the rest. The standard FastQC sections for each sample sit below them. Because the whole experiment is in one report, you compare samples against each other instead of opening a separate file for each one.

Question 4

What are the FastQC checks telling me?

Accepted Answer

FastQC's standard modules cover per-base sequence quality, sequence content, adapter content, duplication levels, overrepresented sequences, and the GC and read-length distributions. Read them as signals rather than verdicts: a quality drop toward the end of the read, heavy adapter content, or a spike in duplicated reads all point to a library that was prepared imperfectly, which is what trimming and the downstream pipeline are there to handle.

Question 5

What do the mitochondrial, ribosomal, and doublet numbers mean?

Accepted Answer

Mitochondrial percentage is the share of reads coming from mitochondrial genes, and a high value often points to stressed or damaged cells. Ribosomal percentage flags libraries dominated by ribosomal RNA, which usually traces back to library preparation. The doublet estimate is a rough count of droplets that likely captured more than one cell. All three come from a downsample of about 5 million reads, so treat them as an early look at library quality rather than a final call.

Question 6

Is there a pass or fail threshold for the QC results?

Accepted Answer

There is no single threshold. QC values are most useful in comparison: a sample with a much higher mitochondrial percentage, a much shorter read length, or a much lower read count than the rest of the batch is worth a second look before a full run. The report is a link rather than a file to email, so your core facility and your bioinformatician open the same view and agree on what it means without another round of questions.

Question 7

How does this output lead into a pathway analysis, if I want one?

Accepted Answer

Not directly, and that is the point: pathway analysis needs a differential expression result, and Quality Control is what sets the next run up to produce one. The recommendations table gives the bulk or single-cell pipeline its CPU and memory settings, chemistry detection configures the single-cell alignment, and the strandedness call shapes the bulk alignment. The bulk pipeline then hands over gene counts, a differential expression run turns those counts into a gene table with fold-changes and adjusted p-values, and that table is the input to a pathway analysis run. Every step in that chain is optional: many experiments stop at QC, others go all the way to pathways.

Question 8

Should I run Quality Control before another pipeline, or on its own?

Accepted Answer

Both work. Running it first hands the downstream pipeline matched CPU and memory settings, a detected chemistry for single-cell data, and a strandedness call for human bulk RNA-seq, which is the cheapest place to catch a problem. Running it on its own gives you a quality record for data you are not analyzing yet, or a check you can hand back to a sequencing provider.

Quality Control

Input

Output

From reads to answers

Why each step is here

FastQC

Resource recommendations

Chemistry detection

Single-cell pre-QC

Strandedness check

Summary tables

MultiQC

What comes back

Quality Control FAQ

Take the next step

Run it in Horizon

Need a tailored pipeline?