Omicser — Omics Engineer

Omics field

Transcriptomics

Transcriptomics is the study of all RNA transcripts produced in a cell, tissue or sample. It measures which genes are transcribed and at what relative level under a defined condition and time point.

Questions this field can answer

  • Which genes change expression between two defined conditions?
  • Which cell types are present in a tissue, and how do their programmes differ?
  • Which isoforms or splice variants are used in a given context?
  • Where in a tissue is a transcript expressed, when spatial methods are used?
  • Do observed differences survive correction for multiple testing and batch structure?

Method

Research workflow

Established practice, from framing a question to depositing data others can reuse.

  1. 01

    Study design

    Fix the contrast first. Use biological replicates, not technical ones — three per group is a floor, not a target — and balance groups across processing batches.

  2. 02

    Sampling

    Standardise harvest time, tissue handling and RNA stabilisation. RNA degrades quickly; record RIN or equivalent integrity scores.

  3. 03

    Assay and data generation

    Choose bulk RNA-seq, single-cell or spatial. Decide poly-A versus total RNA, stranded versus unstranded, read length and depth before sequencing begins.

  4. 04

    Quality control

    Check per-base quality, adapter content, duplication, rRNA fraction, mapping rate and, for single-cell, per-cell counts, gene detection and mitochondrial fraction.

  5. 05

    Analysis

    Align or pseudo-align to a named genome and annotation release, count features, normalize for library size and composition, then model counts with an established negative-binomial framework.

  6. 06

    Interpretation

    Expression is not protein abundance or activity. Report effect sizes with adjusted p-values, and treat enrichment results as hypothesis generating.

  7. 07

    Reproducibility

    Record reference release, tool versions, parameters and the full sample-to-group mapping. Share the count matrix and the analysis code.

  8. 08

    Deposition

    Deposit raw and processed data in GEO or ArrayExpress with a complete sample characteristics table, and cite the series accession.

Practice

Samples, technologies and what can go wrong

Method choice sets the ceiling on what an analysis can show. Limitations are part of the method, not an afterthought.

Sample types

  • Fresh or snap-frozen tissue
  • Cultured cell lines and primary cells
  • Whole blood and PBMCs, where globin depletion matters
  • Dissociated single-cell and single-nucleus suspensions
  • FFPE archival tissue, where RNA is fragmented
  • Tissue sections for spatial transcriptomics

Major technologies

Bulk RNA-seq
Averages expression across all cells in a sample; sensitive and cheap, but blind to cell-type composition shifts.
Single-cell RNA-seq
Profiles individual cells; reveals cell types and states, but is sparse, dropout-prone and sensitive to dissociation stress.
Spatial transcriptomics
Retains tissue location at the cost of resolution or gene panel size, depending on the platform.
Microarrays
Still common in older public data; limited to probed genes with a compressed dynamic range.
qPCR
The standard targeted validation method for a small number of candidate transcripts.

Limitations

  • mRNA level is a poor predictor of protein level for many genes.
  • A snapshot cannot separate transcription rate from RNA stability.
  • Annotation choice changes which features are counted at all.
  • Bulk changes can reflect a shift in cell composition rather than regulation.
  • Single-cell dropout makes absence of a transcript weak evidence.

Common confounders

  • Batch effects from library prep day, flow cell or operator.
  • RNA degradation differing systematically between groups.
  • Sex, age and tissue region unbalanced across conditions.
  • Sequencing depth or cell number differing by group.
  • Dissociation-induced stress genes in single-cell data.

Live data

Search the public record

Read-only searches against public databases. Madomic presents and explains the records; the databases named below remain their source and owner.

Madomic Research Explorer

Search public GEO datasets

Live, read-only search of the GEO DataSets index through NCBI E-utilities, run from the server, capped at eight records and throttled well below the public unauthenticated limit.

Requests are spaced conservatively to respect NCBI's usage policy. Nothing you type is stored.

Reference

Glossary

Essential terms, in plain English.

Series (GSE)
A GEO study record grouping related samples and their processed data.
Sample (GSM)
A single GEO sample record within a series.
Counts
The number of reads assigned to a gene or transcript in a sample.
Normalization
Adjusting counts for library size and composition so samples can be compared.
TPM / FPKM
Within-sample scaled expression units; useful for comparing genes, not for differential testing.
Differential expression
Statistical testing of expression change between defined groups.
FDR
False discovery rate — the expected fraction of false positives among reported hits.
Batch effect
Systematic technical variation aligned with processing, not biology.
Contrast
The precise comparison a model tests, e.g. treated versus control at 24 hours.

For students

Learning path and a practical activity

Everything below uses public data only. No samples, credentials or paid services are needed.

Learning path

  1. 01Learn transcription, splicing and the difference between mRNA and protein levels.
  2. 02Understand library preparation choices and what each excludes.
  3. 03Read a QC report and say which samples you would drop and why.
  4. 04Learn why counts need normalization and why TPM is not a test statistic.
  5. 05Run through a published differential expression tutorial with public counts.
  6. 06Learn multiple-testing correction and how to read a volcano plot honestly.
  7. 07Study one single-cell dataset and its clustering choices.

Activity — read a GEO series and draft a contrast

  1. 01Search GEO in the explorer below for a topic you know, for example “breast cancer RNA-seq”.
  2. 02Open one series and record its accession, organism, platform and sample count.
  3. 03Write out the sample groups from the characteristics table, including replicate numbers.
  4. 04Note the batch information available — submission date, platform, processing notes — and what is missing.
  5. 05Draft one contrast you would test, and state the covariates you would include in the model.
  6. 06Write a validation plan: which two or three genes you would confirm by qPCR or in an independent public series, and why.

Attribution

Research resources

Authoritative public resources. Omicser is independent of each of them and links to the official source.

NCBI GEO

Maintained by NCBI, US National Library of Medicine

Public archive of gene expression and other functional genomics data, browsable by series, sample and platform.

NCBI E-utilities documentation

Maintained by NCBI

The official guide to esearch, esummary and related endpoints, including usage limits — the interface used by the explorer on this page.

Expression Atlas

Maintained by EMBL-EBI

Curated, reanalysed baseline and differential expression across species, tissues and conditions.

Research and education use only. This page is not a clinical diagnosis, medical advice or a laboratory result. Verify every finding in the originating database and in the peer-reviewed literature before drawing conclusions.