More omics worth knowing
Specialised layers that sit between the main ones. Each answers a question the big four cannot, and several are fertile ground for methods development because their tooling is younger.
Epitranscriptomics
More than 170 RNA modifications are known. m6A (N6-methyladenosine), written by METTL3/METTL14, read by YTHDF/YTHDC proteins and erased by FTO/ALKBH5, affects mRNA stability, splicing and translation. Mapping: antibody-based MeRIP-seq/m6A-seq (peaks, ~100–200 nt resolution) and miCLIP (single-nucleotide), antibody-free GLORI and eTAM-seq (quantitative stoichiometry), and direct RNA nanopore sequencing, where modifications alter the current signal (m6Anet, CHEUI, xPore, Dorado modification models). Pseudouridine (BID-seq), m5C (bisulfite on RNA) and A-to-I editing are the other common targets. Gap: quantitative, isoform-resolved maps and validation of nanopore-based calls.
Translatomics
Ribo-seq sequences ribosome-protected fragments (~28–30 nt) and shows where ribosomes are, with 3-nt periodicity revealing the reading frame. Uses: translation efficiency (Ribo-seq / RNA-seq; anota2seq, Xtail, riborex, deltaTE), discovery of upstream and small ORFs and micropeptides (RiboCode, ORFquant, PRICE, RiboTISH, Ribo-TISH), and ribosome stalling. QC with RiboseQC and riboWaltz. GENCODE now catalogues Ribo-seq ORFs (Mudge et al. 2022).
Immunomics and immune repertoires
T- and B-cell receptors are generated by V(D)J recombination, producing ~1015 possible receptors. AIRR-seq sequences these from bulk or single cells (10x VDJ). Tools: MiXCR, IgBLAST, TRUST4 (from RNA-seq), Immcantation (change-O, SHazaM for somatic hypermutation, clonal lineages), scirpy (single-cell, scverse). Questions: clonal expansion, convergent responses to vaccines and infection, antigen specificity prediction (TCRdist, GLIPH2, and deep models with modest generalisation to unseen epitopes). Immunopeptidomics measures HLA-presented peptides by MS to train neoantigen predictors (NetMHCpan, MHCflurry).
Glycomics and glycoproteomics
Glycans are branched, non-templated structures attached to proteins and lipids. They control protein folding, cell recognition, immune function (IgG Fc glycosylation tunes antibody effector activity) and are altered in cancer. Analysis: released-glycan MS or HPLC, intact glycopeptide MS (pGlyco, MSFragger-Glyco, Byonic), databases GlyTouCan and GlyGen. Gap: structural isomers are hard to distinguish, and glycan biosynthesis is only indirectly coded in the genome.
Cell-free DNA, liquid biopsy and fragmentomics
Dying cells release DNA fragments into blood, with a ~167 bp peak reflecting nucleosome protection. Tumour-derived cfDNA (ctDNA) can be detected through mutations (deep targeted sequencing with UMIs), copy number (low-pass WGS; ichorCNA estimates tumour fraction), methylation patterns that also reveal tissue of origin (multi-cancer early detection tests), and fragment size, end motifs and coverage at TSSs (DELFI and related fragmentomics methods). Non-invasive prenatal testing uses the same principles. Gap: sensitivity for early-stage cancers, where ctDNA fractions can be below 0.1%.
Interactomics and functional genomics screens
Pooled CRISPR screens (knockout, CRISPRi, CRISPRa, base and prime editing) measure fitness or reporter effects for every gene or variant; analyse with MAGeCK, BAGEL2, CRISPResso2 (editing outcomes) and the DepMap resources (Chronos-corrected gene effects across ~1,000 cancer cell lines). Perturb-seq combines CRISPR perturbations with single-cell RNA readouts; genome-scale Perturb-seq (Replogle et al. 2022) maps transcriptional consequences of nearly every expressed gene. MPRAs (massively parallel reporter assays) test thousands of regulatory variants for activity at once.
Phenomics, imaging and clinical omics
- Cell Painting: multiplexed fluorescent stains and CellProfiler or deep features produce morphological profiles for drug and gene perturbations (JUMP-Cell Painting consortium).
- Pathomics: deep learning on whole-slide H&E images predicts mutations, expression and survival; foundation models (UNI, CONCH, Virchow, Prov-GigaPath) provide general-purpose embeddings.
- Radiogenomics: linking imaging features to molecular subtypes.
- EHR phenomics: phenotype definitions (phecodes, HPO, OMOP common data model) enable PheWAS and biobank-scale genetics.
Exposomics
The exposome is the totality of environmental exposures over a lifetime: chemicals, diet, pollutants, infections, social factors. High-resolution MS of blood and urine detects thousands of exogenous chemicals; ExWAS (exposome-wide association studies) parallel GWAS. Gaps: chemical identification (even darker than the endogenous metabolome), time-varying exposure, and correlated exposures that confound each other.