What RNA-Seq Measures

RNA-Seq measures gene expression by sequencing the RNA that a sample contains at the moment it is collected. Rather than inferring activity indirectly, you count transcript-derived molecules directly, so the data reflect what the cells were actually transcribing. This lesson traces the full path from a biological sample to a differential-expression result so the rest of the course has a map.

From RNA To Counts

In the lab, RNA is extracted, fragmented, and reverse-transcribed into complementary DNA (cDNA), which is sequenced to produce millions of short reads. Each read is a fragment of some transcript, so aligning reads back to their gene of origin and tallying them estimates how much of each transcript was present. Because more reads land on more abundant transcripts, the read count per gene is a proxy for that gene's expression level.

Older microarrays measured expression by hybridizing labeled cDNA to a fixed set of probes, so they could only report on genes designed onto the chip and saturated at high signal. RNA-Seq is open-ended: it can detect novel transcripts, splice isoforms, and a far wider dynamic range because it counts discrete reads instead of reading a fluorescence intensity. That is why bulk RNA-Seq has largely replaced arrays for routine expression profiling.

The Analysis Workflow

Almost every bulk RNA-Seq project follows the same backbone regardless of organism. You begin with raw reads, check their quality, align or quantify them against a reference, collapse those results into a count matrix, and finally test for expression differences between conditions.

FASTQ reads
  -> quality control    FastQC, MultiQC
  -> align or quantify  STAR, HISAT2, or Salmon
  -> count per gene     featureCounts, or Salmon quant
  -> count matrix       genes x samples
  -> diff expression    DESeq2, edgeR

Prep, Depth, Read Ends

Library prep decides which RNA you actually sequence: poly-A selection captures mature mRNA, while ribosomal RNA depletion keeps a broader pool that includes many non-coding RNAs. Sequencing depth, the number of reads per sample, sets how confidently you can quantify lowly expressed genes, with bulk expression studies commonly targeting roughly 20 to 30 million reads per sample. Reads are produced single-end (one read per fragment) or paired-end (both ends of each fragment), and paired-end reads improve mapping across splice junctions and repetitive regions.

# A FASTQ file stores 4 lines per read; count reads in the R1 (_1) file of a paired-end run
zcat SRR1039508_1.fastq.gz | wc -l | awk '{print $1/4, "reads"}'

22935521 reads