IV. Ecosmomics: Independent, UniVersal, Complex Network Systems and a Genetic Code-Script Source

2. The Innate Affinity of Genomes, Proteomes and Language

Flam-Sherperd, Daniel, et al. Atom-by-atom protein generation and beyond with language models. arXiv:2308.09482.
We post an August entry by University of Toronto and Vector Institute reseachers including Alán Aspuru-Guzik to record much current activity in biocomputional studies which now join Large Language Models of AI neural machine learning methods. As the excerpt cites, a broad continuity across chemical, genetic, biochemical and onto linguistic phases bodes for an innately informational, universe to wumanverse, literacy to literacy procreative milieu. See also, for example, PEvoLM: Protein Sequence Evolutionary Information Language Model by Issar Arab at 2308.08578.

Protein language models learn powerful representations directly from sequences of amino acids. In contrast, chemical language models learn atom-level results of smaller molecules that include every atom, bond, and ring. In this work, we show that chemical language models can learn atom-level proteins which can generate the standard genetic code and far beyond it. The results demonstrate the potential for biomolecular design at the atom level using language models. (Exerpt)

Gimona, Mario. Protein Linguistics and the Modular Code of the Cytoskeleton. Barbieri, Marcello, ed. The Codes of Life. Berlin: Springer, 2008. The University of Salzburg geneticist contributes to the long project to interpret, join and unify the molecular and literal versions, in support of the growing conclusion that “Nature is Structured in a Language-like Fashion.” See also an earlier paper “Protein Linguistics – A Grammar for Modular Protein Assembly?” in Nature Reviews: Molecular Cell Biology (7/1, 2006).

Hackenberg, Michael, et al. Clustering of DNA Words and Biological Function: A Proof of Principle. Journal of Theoretical Biology. 297/127, 2012. University of Granada and University of Malaga, Spain system biologists including Pedro Carpena contribute to historic 2010s verifications that the molecular nucleotide version and human cultural literature are one and the same, that they are formed and suffused by the same informative nonlinear complex network systems. View articles of this kind, for example, in the journal Complexity over recent years.

Relevant words in literary texts (key words) are known to be clustered, while common words are randomly distributed. Given the clustered distribution of many functional genome elements, we hypothesize that the biological text per excellence, the DNA sequence, might behave in the same way: k-length words (k-mers) with a clear function may be spatially clustered along the one-dimensional chromosome sequence, while less-important, non-functional words may be randomly distributed. To explore this linguistic analogy, we calculate a clustering coefficient for each k-mer (k=2–9 bp) in human and mouse chromosome sequences, then checking if clustered words are enriched in the functional part of the genome. The clustering of DNA words thus appears as a novel principle to detect functionality in genome sequences. As evolutionary conservation is not a prerequisite, the proof of principle described here may open new ways to detect species-specific functional DNA sequences and the improvement of gene and promoter predictions, thus contributing to the quest for function in the genome. (Abstract excerpt)

Heckmeier, Philipp, et al.. A billion years of evolution manifest in nanosecond protein dynamics. PNAS. 121/10, 2024. We cite this paper by University of Zurich and Columbia University biochemists as an example of how far the scope and range of these current techniques can reach. And again who are we peoples with an Earthomo sapience to be able to look down and back and reconstruct and re-present how it all came to occur?

Protein dynamics forms a broad bridge between structure and function, yet the impact of evolution on ultrafast protein processes remains enigmatic. This study delves into the nanosecond-scale phenomena of a conserved protein across species separated by almost a billion years as a way to investigate ten complex homologs. In so doing, we found a cascade of rearrangements which manifest in discrete time points over hundreds of millions of years. Our work poses a novel scientific inquiry within molecular paleontology compared by the rapid pace of protein processes which can connect the shortest time scale in living matter (10^-9 s) with the largest ones (10^16 s). (Abstract)

Holzer, Jacqueline. Genomes & Language. http://www.liu.se/isk/research/doc/Birgitta_forum.pdf. An extensive summary from a Birgitta Forum held in August 2002 in Vadstena, Sweden, reviewed more in Emergent Genetic Information.

Holzer, Jacqueline. Genomes & Language. http://www.liu.se/isk/research/doc/Birgitta_forum.pdf. A website for the conference program and lengthy Concluding Reflections from a Birgitta Forum held in August 2002 in Vadstena, Sweden. Geneticists and linguists are finding much commonality between these archetypal formative modes upon which our life and world is founded. A main resource is the work of the German philosopher Wolfgang Raible, who also spoke, Google for his 2001 paper “Linguistic and Genetics. Systematic Parallels”.

Geneticists, when presenting the structure of the human genome, seem to find the metaphor of the genome as a book, or a text, useful. Genomes and texts are both multiply articulated structures, where purely contrastive units – phonemes, letters, bases – combine to form meaningful units at several levels of increasing complexity – words, sentences, texts; codons, genes, chromosomes. (4) In a very profound way he (Raible) shows the structural similarities between linguistics and genetics and sees herein a “deeper relationship between the ‘grammar of biology’ and the grammar of natural languages.” In both systems, the principles allowing the reconstruction of multi-dimensional wholes from linear sequences of basic elements are identical: double articulation, different classes of ‘signs,’ hierarchy, combinatorial rules: wholes are always more that the sum of their parts. (Holzer, 5)

Hwang, Yunha, et al. Genomic language model predicts protein co-regulation and function. Nature Communications.. 15/2880, 2024. We enter this work by Cornell, Harvard, Johns Hopkins, and MIT biologists including Sergey Ovchinnikov as another literate version of the textual affinity of nucleotides and narratives. See also ProteinEngine: Empower LLM with Domain Knowledge for Protein Engineering at arXiv:2405.06658.

•
Deciphering the relationship between a gene and its genomic context is vital to understand and modify biological systems. Machine learning can study the sequence-structure-function paradigm but higher order genomic information remains elusive. Evolutionary processes dictate genomic contexts in which a gene occurs across phylogenetic distances, and these emergent patterns can be leveraged to uncover functional relationships. Here, we train a genomic language model (gLM) on metagenomic scaffolds to uncover regulatory relationships between genes. Our findings illustrate that gLM’s deep learning of metagenomes is an effective approach to encode the semantics and syntax of genes and uncover complex relationships in a genomic region. (Abstract)

The unprecedented amount and diversity of metagenomic data presents opportunities to learn hidden patterns and structures of biological systems. With larger amounts of data, these models can disentangle the complexity of organismal genomes and their encoded functions. The work presented here validates the concept of genomic language modeling. Our implementation of the masked genomic language modeling illustrates the feasibility of training such a model, and provides evidence that biologically meaningful information is being captured in learned contextualized embeddings. (9)

Igamberdiev, Abir and Nikita Shklovskiy-Kordi. Computational Power and Generative Capacity of Genetic Systems. BioSystems. 142-143/1, 2016. A Memorial University of Newfoundland theoretical biologist and a National Research Center for Hematology, Moscow research physician contribute to the intent of this journal (second quote) to achieve a natural philosophy of life’s evolution as an oriented ascent from an innately conducive cosmos. In this encompassing genesis, a “generative” agency is a textual essence which rises in kind from a physical matrix to genomic and linguistic manifestations. Once again, after decades of study, it is strongly put that these two prime codes are one and the same.

In his many writings, AI cites Aristotelian, Greek, and Renaissance roots to provide a historical heritage for this 21st century resolve. In this paper it is said that Heraclitus’ “self-growing Logos” can now be confirmed. See his website at www.mun.ca/biology/igamberdiev/index.php for a publications page, such as Relational Universe of Leibniz (2105), and Semiotic Autopoiesis of the Universe (2001). A “quantum measurement” theory is broached here, explained more elsewhere, which means that biological systems survive, evolve, and prosper by recursively comparing new experience with prior experiential representations.

Semiotic characteristics of genetic sequences are based on the general principles of linguistics formulated by Ferdinand de Saussure, such as the arbitrariness of sign and the linear nature of the signifier. Besides these semiotic features that are attributable to the basic structure of the genetic code, the principle of generativity of genetic language is important for understanding biological transformations. The problem of generativity in genetic systems arises to a possibility of different interpretations of genetic texts, and corresponds to what Alexander von Humboldt called “the infinite use of finite means”. These interpretations appear in the individual development as the spatiotemporal sequences of realizations of different textual meanings, as well as the emergence of hyper-textual statements about the text itself, which underlies the process of biological evolution. These interpretations are accomplished at the level of the readout of genetic texts by the structures, which includes DNA, RNA and the corresponding enzymes operating with molecular addresses. The molecular computer performs physically manifested mathematical operations and possesses both reading and writing capacities. Generativity paradoxically resides in the biological computational system as a possibility to incorporate meta-statements about the system, and thus establishes the internal capacity for its evolution. (Abstract)

Life is a self-organizing and self-generating activity of open non-equilibrium systems determined by their internal semiotic structure. My vision of life in the Universe is based on the principles of the quantum measurement theory, which can be considered as a mirrored image of theoretical biology. Life by its existence in self-reflecting loops establishes basic physical parameters of the Universe. The philosophical background of this approach is in Greek philosophy, in monadology of Leibniz, and in the organism philosophy of A. N. Whitehead. The field of theoretical biology is a description of systems that possess their own embedded description. Life always maintains and solves these paradoxes since living organisms possess their internal description. The structure of the Universe includes a self-reflective loop to be observable, i.e. existing. (AI site Theoretical Biology)

BioSystems encourages experimental, computational, and theoretical articles that link biology, evolutionary thinking, and the information processing sciences. The link areas form a circle that encompasses the fundamental nature of biological information processing, computational modeling of complex biological systems, evolutionary models of computation, the application of biological principles to the design of novel computing systems, and the use of biomolecular materials to synthesize artificial systems that capture essential principles of natural biological information processing.

Jolma, Arttu, et al. DNA-dependent Formation of Transcription Factor Pairs Alters Their Binding Specificity. Nature. 527/384, 2015. A Karolinska Institute, Sweden group and colleagues, led by Jussi Talpale, report a unique parsing of nucleotide genetics by treating them much as a linguistic script. The achievement was noted in a Science Daily item for November 15, 2015 (Google SD and article keywords) entitled Complex Grammar of the Genomic Language. A gene regulatory code is thus composed by “DNA words,” which can be seen to combine and compound just as lexicons and sentences.

Karollus, Alexander, et al. Species-aware DNA language models capture regulatory elements and their evolution. Genome Biology.. Vol. 25/Art 83, 2024. In this BMC journal, Technical University of Munich geneticists introduce an effective synthesis of these premier nucleotide and narrative code-script domains. By so doing, a cross-assimilation is achieved of these biomolecular and linguistic text phases to an extent they can be seen as the same descriptive process in different sequential venues. See also How do Large Language Models understand Genes and Cells Chen Fang, et al in bioRxiv preprints for March 27, 2024 and Gene and RNA Editing at arXiv:2409.09057.

Large-scale multi-species genome sequencing promises to shed new light on gene regulatory instructions. To this end, algorithms are needed that can leverage conservation while accounting for their evolution. Here, we introduce species-aware DNA language models trained on 800 species spanning 500 million years of evolution. We show that DNA language models distinguish transcription factor and RNA-binding protein motifs from background non-coding sequence. These results show that species-aware DNA language models are a powerful, flexible, and scalable tool to integrate information from large compendia of highly diverged genomes. (Abstract)

A typical eukaryotic genome contains large regions of non-coding DNA. Tese are not translated into proteins but contain regulatory elements which control gene expression in response to environmental cues. Finding these regulatory elements and elucidating how their combinations and arrangements determine gene expression is a major goal of genomics research and is of great utility for synthetic biology and personalized medicine. (1)

ConclusionIn this study, we trained language models on the genomes of hundreds of fungal species, spanning more than 500 million years of evolution. We specifically directed our attention to non-coding regions, examining the ability of the models to acquire meaningful species-specific and shared regulatory attributes when trained on the genomes of many species. To our knowledge, we are the first to show that LMs are able to transfer these attributes to unseen species.

Kay, Lily. A Book of Life?: How the Genome Became an Information System and DNA a Language. Perspectives in Biology and Medicine. 41/4, 1998. The late philosopher of science discerns intrinsic congruities between the verbal and genetic codes.

Kay, Lily. Who Wrote the Book of Life? Stanford, CA: Stanford University Press, 2000. A premier history of science study of how a linguistic metaphor came to represent the genetic code. The author goes on to note a correspondence between molecular genetics, language and the Chinese divination system, the I Ching.

As with (linguist Roman) Jakobson, the answer was affirmative (to the question of one basic code) and pointed to a universe fundamentally different from that portrayed in Jacques Monod’s Chance and Necessity. Rather than viewing DNA-based life as a product of chance, it would be chance subject to the structures and patterns of the I Ching. And rather than being a gypsy living on the edge of an alien world, as Monod decried, a human being would enjoy a deep sense of security that emerged from being planted physically and spiritually in an internal natural order. (318)

Previous   1 | 2 | 3 | 4 | 5 | 6 | 7  Next