21.3 Beyond Sequence

303

To preserve spatial information in a tissue, in situ hybridization can be used to

obtain transcriptomic information. ⁷ This is called single-molecule fluorescence in

situ hybridization (smFISH), and many variants are being developed.

21.2

Applications to Disease and Other Phenomena

A good illustration of the power of multiomics is the decoding of the molecular pro-

cesses underlying circadian rhythms in green algae. ⁸ Understanding these processes

would scarcely have been possible without the parallel examination of genomic,

transcriptomic of metabolic data. Sharifi-Noghabi et al.2019 have developed useful

techniques for integrating such data. ⁹

The chromosomal disorder characteristic of cancer (Sect. 14.5) makes single-cell

examination of tumours an especially valuable route to obtaining insight into their

formation and development.

A tensor-based association test seems to be a useful for discovering novel disease

genes or mechanisms from multiomics data. ¹⁰

21.3

Beyond Sequence

Proteomics data (see Chaps. 18 and 23) are integrated with sequence information in

the attempt to assign function. Proteins whose mRNA levels are correlated with each

other, proteins whose homologues are fused into a single gene in some organisms,

those which have evolved in a correlated fashion, those whose homologues operate

together in a metabolic path or that are known to physically interact can all be con-

sidered to be linked in some way; for example, a protein of unknown function whose

expression profile (see Footnote 7 in Chap. 18) matches that of a protein of known

function in another organism is assigned the same function. In a literary analogy,

one could rank the frequencies of words in an unknown and known language and

assign the same meanings to the same ranks. Whether the syntax of gene expression

is sufficiently shared by all organisms to allow this to be done reliably is still an open

question. One can, however, probably assume that this syntax is shared by all cells

from the same organism.

Other kinds of data assisting protein function prediction are structure prediction,

intracellular localization, signal peptide cleavage sites of secreted proteins, glyco-

sylation sites, lipidation sites, phosphorylation sites, other sites for posttranslational

modification, cofactor binding sites, dehydron density, and so on.

7 Lubeck and Cai (2012).

8 Strenkert et al. (2019).

9 See also Lee et al. (2020) for a comprehensive review.

10 Chang et al. (2001).