Analysis settings contains Comparison, Read processing, Output options and Advanced tabs. Inactive controls follow the input route: a count table does not undergo trimming, and imported results do not fit a local differential-expression engine.
1. Choose read processing#
For FASTQ/public-read routes, keep the documented fastp trimming default unless your protocol requires another supported trimmer or the input is already trimmed. Raw and post-trimming FastQC reports feed MultiQC.
Core controls#
| Setting | What it changes | When to change it / example |
|---|---|---|
| Trimming: on; fastp default | Removes adapter/low-quality sequence before alignment. Trim Galore and Trimmomatic are alternatives. | Disable only when intentionally using already-trimmed reads. Keep the upstream trimming record. |
| rRNA filtering: off; SortMeRNA selected | Removes reads classified as ribosomal RNA before alignment when enabled. | Enable to implement a planned rRNA-removal step. SortMeRNA uses a database; RiboDetector is a CPU classifier without that reference database. |
| FastQ Screen: off | Reports matches to an existing genome panel; it is a screen, not a contamination-removal filter. | Enable with a configured panel to inspect unexpected sources. No genome panel is downloaded automatically. |
| Aligner: STAR | STAR and HISAT2 provide genome BAMs; Salmon performs transcriptome quantification without BAMs. | Consider HISAT2 or Salmon when resource constraints and the analysis goal warrant them. Changing the route can change estimated counts. |
| Quantifier: featureCounts | Counts annotated features from genome alignments. STAR gene-counts is an alternative with STAR; HISAT2 uses featureCounts and Salmon uses tximport. | Use STAR gene-counts only when you intentionally choose STAR’s counting definition. |
| RSeQC: off | Adds read-distribution and gene-body coverage diagnostics from BAMs. | Enable on STAR/HISAT2 when you need those views; it cannot run on Salmon output. |
| Organellar genes: keep | Keep includes them; discard removes them before DE; separate runs nuclear DE and writes organellar counts/fractions. | For a planned nuclear-only comparison, choose discard or separate and record the choice. Separate does not promise a separate organellar DE fit. |
2. Understand advanced read parameters#
Advanced: trimming settings with examples
| Setting | What it changes | When to change it / example |
|---|---|---|
| fastp quality Phred 15; low-quality limit 40% | Controls the quality classification and tolerated fraction of low-quality bases. | Raise stringency only with a protocol/QC reason; stricter filtering can lose usable reads as well as poor reads. |
| Minimum length 36 | Rejects reads shorter than the retained-length requirement. | Illustration: a 30-base read fails 36. Raising the limit to 50 also removes a 45-base read. |
| Poly-G off; poly-X off | Enables terminal homopolymer trimming. | Use when library/instrument characteristics justify it, not to make a QC panel appear cleaner. |
| Trimmomatic window 4, quality 15; leading/trailing 3 | Sets sliding-window and end-quality trimming. | A stricter window criterion can shorten more reads; preserve the reason for the change. |
| RiboDetector ensure=norrna; chunk size 256 × 1024 reads | Sets classification retention mode and processing batch size. | Change chunk size for memory constraints; it is not a biological significance threshold. |
| FastQ Screen subset 100000; panel config required | Controls how many reads are sampled and which genomes are screened. | A larger screen samples more reads but costs time; an absent organism in the panel cannot be detected by that panel. |
Advanced: alignment and counting settings
| Setting | What it changes | When to change it / example |
|---|---|---|
| STAR two-pass off | Controls a second alignment pass informed by detected splice junctions. | Enable for an intentional alignment protocol, then treat outputs as a changed analysis. |
| STAR maximum multimappers 10; mismatch/read-length ratio 1.0 | Constrains reported mapping multiplicity and mismatch acceptance. | Tighter constraints may reduce assigned reads; do not tune solely to increase a downstream hit count. |
| featureCounts feature exon; attribute gene_id | Defines the annotation features and grouping key counted as genes. | A custom annotation must actually contain the chosen fields. Changing exon to another feature changes the quantified object. |
Strandedness is inferred and applied per sample in the published baseline. Inspect results/aligned/strandedness_per_sample.tsv; a configured generic code is not evidence that every library has that orientation.
3. Save the settings#
Click Save analysis settings. Complete Comparison, then choose resources and run pre-run checks.
Terms on this page: Strandedness.