Before importing, establish whether your file contains reads, raw counts, normalized microarray intensities or results from a fitted model. Similar-looking tables can represent different quantities.
Five ways in#
| Setting | What it changes | When to change it / example |
|---|---|---|
| Public accessions | Fetches public metadata and reads, then runs the read-processing route. | Use SRR / SRP / PRJ accessions, or an RNA-seq GSE resolved to SRA runs. |
| Local FASTQ | Skips public download; retains read QC, alignment/quantification and modelling. | Use when reads are already on disk. One run must have one read layout. |
| Count matrix | Skips download, read QC, alignment and counting; fits a count model. | Use raw unnormalized counts, not TPM, FPKM or log-CPM. |
| Imported differential-expression results | Skips local count modelling and uses the supplied statistics. | Confirm the source direction; see imported results. |
| Microarray | Uses processed intensities or supported Affymetrix CEL processing, then limma. | Use the Microarray tab, not the raw-count importer. |
1. Select the route on Add data#
The route tabs are Public accessions, Local FASTQ, Count / DE tables and Microarray. The Count / DE tables tab contains two distinct import buttons.
2. Add public or local reads#
For public data, paste accessions and click Fetch metadata & build samples. Review the inferred layout, FASTQ URLs, read counts and suggested condition labels. For local reads, click Select FASTQ Files and verify that the first and second mates belong to the same biological sample. Missing or swapped mates are not repaired by an aligner setting.
3. Import counts when counting is complete#
Click Import count matrix. Supply a gene-by-sample table, with sample columns that can be matched to the sample sheet. Every cell must be present, numeric, finite and non-negative. Raw integer counts are the default requirement. Any non-integer value triggers the explicit RSEM/tximport estimated-count prompt; accepted estimates are rounded with round-half-to-even. This opt-in is not permission to import normalized abundances.
4. Verify the imported samples#
Open Samples, inspect every row, save the sheet and select the organism on Reference. For intensities follow Microarray input; for finished statistics follow Imported results.
Troubleshooting: a table is rejected
Check delimiters, unique gene identifiers, missing cells, finite non-negative numeric values and sample-column names. Do not suppress a format error by changing the biological meaning of the data. Export a plain TSV/CSV from the upstream tool and retain its original file and method record.
Terms on this page: FASTQ · Count matrix.