Journal of Data Mining in Genomics & Proteomics

A General Evolution Landscape of Language and Cognition Genes

Zhizhou Zhang^1*, Shuaiyu Zhang², Hongjie Zhou² and Yongdong Xu^{2* 1}BIOX Center for Marine Biotechnology, Harbin Institute of Technology, Weihai, China
²School of Computer Science and Information Technology, Harbin Institute of Technology, Weihai, China
^*Correspondence: Zhizhou Zhang, BIOX Center for Marine Biotechnology, Harbin Institute of Technology, Weihai, China, Email: Yongdong Xu, School of Computer Science and Information Technology, Harbin Institute of Technology, Weihai, China, Email:

Received: 27-Feb-2024, Manuscript No. JDMGP-24-25034; Editor assigned: 01-Mar-2024, Pre QC No. JDMGP-24-25034 (PQ); Reviewed: 15-Mar-2024, QC No. JDMGP-24-25034; Revised: 22-Mar-2024, Manuscript No. JDMGP-24-25034 (R); Published: 29-Mar-2024, DOI: 10.4172/2153-0602.24.15.339

Introduction

There are various, yet complementary, theories about the evolution of human language [1-6]. Clearly, all animals have their own ways of communication, even if not always through vocal sounds made with the mouth. Primates, for example, have at least a dozen distinct vocal sounds that carry specific meanings, which can be considered a basic form of language. Some studies strongly believe in the following hypothesis: once the primate brain evolved to a certain stage and suddenly acquired symbolic thinking ability, Homo sapiens came into being [5,6]. Symbolic thinking naturally possesses the capacity to gradually refine and complicate the meanings of language. Once language could become more complex and precise in meaning, and its significance could be passed down, humans were able to progressively accumulate their ancestor’s experiences, accelerating human evolution.

The above process also implies several facts: 1) the evolution of language occurred much earlier than the ability for symbolic thinking. The capacity for language is primarily a capability of vocal sounds, resulting from the movement of muscles inside the body such as those in the mouth, and is fundamentally a type of motor skill; (2) Symbolic thinking is not a uniquely human cognitive ability, but there are varying degrees of symbolic thinking. Ordinary animals, even if they appear relatively intelligent, cannot compare with human symbolic thinking. If we disregard the levels of symbolic thinking and simply measure its presence or absence, then this point in time likely occurred about 35,000~70,000 years ago [5,7,8]. If we use different levels to measure symbolic thinking, we can better understand why human evolution is divided into stages such as ancient apes, hominids, Homo erectus, Homo habilis, Homo sapiens, and modern humans; (3) Language and cognitive abilities complement each other, so the period following the emergence of Homo sapiens in the evolutionary process (which could span tens of thousands to hundreds of thousands of years) should have been a time of rapid advancement in both language and cognitive abilities. Through the proliferation, interbreeding, and iteration of populations, most human groups on Earth gradually came to possess advanced language and cognitive abilities, while those groups and individuals with only one of these advanced abilities would accelerate towards extinction; (4) Due to the vast and diverse geographical environments on Earth, various conditions are provided for the long-term existence of specific populations, so there is still, in principle, genetic diversity in language and cognitive abilities on Earth. Even in large cities, or to say, around us, we can still encounter individuals with severe language impairments but outstanding cognitive abilities. There are also many individuals who are proficient in spoken language but have particularly weak symbolic thinking abilities (such as in mathematics). Some children nearing the age of 10 still do not possess adequate language skills. These are probably all specific intermediate states in the evolutionary process of the two abilities mentioned above; (5) The nature of language determines that it only has clear meanings within specific contexts, which aligns with the current situation where there are over 7,500 languages worldwide. However, human evolution has far surpassed these limitations. Humans can now design entirely new languages independent of any local context and teach and spread them anywhere; (6) As the brain’s spatial structure increasingly shows clear correspondence and working principles for language and cognitive abilities, detailed quantitative measurement of these abilities in principle allows for definitive diagnosis of brain diseases, aging, etc., and forms the basis for developing new brain- computer interface technologies and products.

The process of language evolution and cognitive evolution should be reflected in the human genome sequence, and the modern human genome should also contain some of this evolutionary information. The older the fossil DNA, the more likely it is to reveal a greater number of intermediate states of evolution. Moreover, in principle, one should be able to observe different evolutionary rhythms for language genes and cognition genes along the evolutionary path. This study attempts to analyze a set of linguistic gene polymorphisms and cognitive gene polymorphisms in a batch of ancient DNA and modern genome sequences, thereby depicting a general landscape of the evolution of language and cognition genes.

Materials and Methods

Genome sequences

Genome sequences were downloaded from ENA database (https://www.ebi.ac.uk/ena/browser/), SRA database (https://www.ncbi.nlm.nih.gov/sra) and Ensembl genome browser. Total 170 whole genomes (including 59 ancient genomes) from 5 continents (Africa, Asia, Europe, North America, and South America) were collected. The above genome sequences have fastq, fn or fna formats, and all can be read and scanned with python based hash07plus03 software (Table 1).

Table 1: The 170 whole genomes employed in this study.

Language/Cognition genes and their SNPs

Language is an emergent complicated function of human being, though many other animals also have their own ‘Languages’. If a gene mutation is statistically or experimentally associated with a certain language function loss, it would be called language gene. For both language gene and cognition gene, SNP sites in the dbSNP database were selected in a way that the each whole gene region was relatively, equally spanned by the selected sites, plus those already with known clinical effects (seen in the Genecards database). Table 2, listed 36 language/cognition genes, and a total 239 SNPs from 18 language genes were selected for this study (Table 3), while 223 SNPs from 18 cognition genes were selected (Table 4).

Table 2: Selected language/Cognition genes in this study [9-15].

Table 3: Tested 239 SNPs of 18 language genes.

Table 4: Tested 223 SNPs of 18 cognition genes.

Genome sequence analysis software development

The SNP (Single Nucleotide Polymorphism) loci finding software based on hash tables primarily processes biological whole genome files and rapidly identifies SNP loci within the genome by utilizing a hash table-based search algorithm, obtaining the specific values of the mutated bases at these loci (A/C/G/T). The software is written in Python. Initially, it processes three different formats of whole-genome files—fastq, fna, and fa-based on their unique characteristics, extracting gene sequences and generating standard format files that include all lines containing only ATCGN bases. Subsequently, it reads the SNP file and stores the information for each locus into a hash table. The target sequence is converted into binary representation according to the corresponding rules of A-00, T-01, C-10, G-11, and then into decimal form to serve as keys in the hash table. For the standard files, the software reads sequences in groups based on the matching length and converts them into the corresponding decimal representation for comparison with the keys in the hash table. If a match is successful and the number of matched bases does not exceed the limit, the matched bases and location information are added to the corresponding value. This process is repeated until the end of the file. Upon completion of each file match, the results are tallied to determine the matched bases and their respective quantities for each SNP locus and the findings are produced. During use, the software can process multiple genome files in batches and impose restrictions on the matching length and the number of matches for a single SNP. After extensive validation, this software has shown a significant improvement in speed compared to conventional matching algorithms and other software based on KMP (Knuth–Morris–Pratt) improved algorithms.

Sample SNP information abstraction and PCA analysis

The authors used 010 Editor Software to extract SNP information from genome files, but most SNP information was abstracted with hash07plus03 software. In all 170 genomes, the sizes mainly range from 200 M to 120 G. Genomes with fastq format but less than 10 G were generally neglected or only used as a reference. Principal Component Analysis (PCA) was performed using R packages FactoMineR, factoextra, ggrepel and ggplot2. R codes for PCA [16].

SNP profile similarity measurement among samples

In order to compare the similarities in SNP profiles between the samples of each genome file, we used three suitable similarity calculation methods to conduct similarity analysis on the SNP site bases matched by all samples, and obtained the following results.

Method 1: Levenshtein distance algorithm: This method measures the difference between bases at two sites by calculating the edit distance, that is, the minimum number of operations (insertions, deletions, or substitutions) required to convert bases at one site into bases at another site. We combine all the editing distance of the bases at the site is summed and divided by the maximum number of bases in the calculated sample to obtain the difference rate, which is converted into similarity [17].

Method 2: Smith-Waterman algorithm: This method uses dynamic programming to find the local optimal alignment between two sequences, that is, the subsequence with the highest similarity between the two sequences. We united the bases of all sites into a string, then used this algorithm to calculate the similarity score between the two strings, and then normalized the scores to obtain the similarity rate [18].

Method 3: Needleman-Wunsch algorithm: This method is similar to method 2, but this method adopts a global optimal alignment scheme, that is, considering the overall similarity of the two sequences instead of the local similarity. We also unite the bases of all sites into a string, and then use this algorithm to construct a score matrix to calculate the similarity score between the two strings, and then obtain the similarity rate [19]. We applied these three methods to the calculation of similarity, and obtained the similarity between any two samples, as well as the similarity of all samples relative to the reference sample p6, and also drew a similarity change curve to intuitively demonstrate changing trends.

Results and Discussion

The PCA analysis results (Figures 1-3), show that among all samples, various lower animals are situated at the farthest left position, followed closely by a batch of ancient human samples. The genomic sequences of all animals are whole-genome sequences, whereas many ancient human genome sequences are incomplete, such as c20 and nd10, which only have 1.7G and 200M respectively. Therefore, the clustering of these ancient human samples with lower animals can only indicate that certain SNP patterns in these ancient human samples belong to a more primitive state, corresponding to the initial evolutionary stage of ancient humans. Modern humans are mostly located at the farthest right position in the figures, so from left to right, it essentially reflects the evolutionary stages from low to high. Interestingly, samples corresponding to each evolutionary stage seem to simultaneously include origins from Asia, Africa, Europe, and the Americas. This suggests that new evolutionary populations generally had enough time to spread across nearly all continents; with the current sample size, it appears that European samples, especially Neanderthal samples, have the highest occurrence rate in the earliest evolutionary stages of language genes and cognitive genes.

Figure 1: CG PCA result for SNPs in cognition genes. Note: Other A; South America.

Figure 2: LG PCA result for SNPs in language genes. Note: Other A; South America.

Figure 3: CGLG PCA result for SNPs in language and cognition genes. Note: Other A; South America.

The evolutionary process of language genes (Figure 2) differs significantly from that of cognitive genes (Figure 1). The early patterns of SNP diversity in language genes are very similar across all animal samples, whereas cognitive genes exhibit significant variation right from the start among the selected animal samples. There are two polymorphic states of cognitive genes at the beginning of evolution, one state is reflected in a group of intelligent animals (such as border collies, camels, alpacas, cetaceans, cheetahs, etc.) and primates, while the other state is directly reflected in certain animals and early human populations, and these two states gradually diverge and then come close again, but they never overlap; the closest distance between these two states is reflected in the orangutan species, because the polymorphic patterns of cognitive and language genes in chimpanzees and bonobos are closest to those of modern humans.

The Asian sample c20 (Tianyuan) consistently appears at the earliest stage of evolution. Since the whole-genome sequence of this sample is not complete, the conclusions drawn from it are currently insufficient, but it generally tends to appear with European Neanderthals at the earliest stages of evolution, which at least suggests that some SNP loci of this sample may belong to the most ancient combination patterns. The Tianyuan genome possesses genomic features close to modern Asians, who carry approximately 4%-5% Neanderthal DNA shared by Upper Paleolithic Eurasians. The Tianyuan genome has a relatively closer relationship with present-day and ancient Asians than with Europeans [20]. European sample nd10 shares a similar situation with the Asian sample c20, that is, due to the extremely incomplete genome sequence, the results obtained lack sufficient credibility. The only reference worth mentioning is that some of its SNP patterns are in the oldest state of evolution. Clearly (Figures 1-3), it is not feasible to discern whether the oldest SNP patterns originated from Asia, Africa, or Europe.

Comparing the evolutionary process of language genes and cognitive genes in European samples, the basic conclusion is that Neanderthals have consistently been at the early stages of evolution. Among all African samples, the Moroccan sample seems to be at the earliest stage of evolution, and indeed there is literature supporting that the earliest Homo sapiens from Africa originated from the Moroccan region [21,22].

Quasi-quantitative measurement among CG/LG/CGLG SNP profiles for 170 samples is essential to observe the potential points at which big changes happen. SNP profile similarity measurement results may tell something (Figure 4). In Figure 4, there are roughly two points (see two arrows) showing significant value changes for three different measurements (Levenshtein Distance, Smith- Waterman similarity and Needleman-Wunsch similarity; those with three measurements significantly changed at the same time can be regarded as turning points), and around the two points, there are total about 14 samples (ne9m, et1, nd5n, sa3, us2, ge1, ss3, ss2, c19, ne3, c16, ne2, c24, c15). All these 14 samples are located at the big arrow position in Figure 1. In Figure 5, there are also roughly two turning points (see two arrows), around which there are basically 9 samples (c5, c26, c7, c8, mg1, nd5n, us2, ge1, et1). In Figure 6, there are another two turning points that harbor 5 samples (us2, ge1, ja1, c19 and nd8). The Figure 4, demonstrated that CG and LG plus CGLG profiles have 2 common samples (ge1 and us2) at their turning points, suggesting that Europe or North America might be also key sites for human evolution, at which sites some critical changes in both language development and cognition function especially symbolic thinking may experience an accumulative jump.

Figure 4: The similarity of SNP profiles among all samples in CG relative to the p6 sample. Note: Levenshtein distance; Needleman_Wunsch.

Figure 5: The similarity of SNP profiles among all samples in LG relative to the p6 sample. Note: Levenshtein distance; Needleman_Wunsch.

Interestingly, primate samples can be observed at the aforementioned turning points, and the extent of their presence varies across the turning points in Figures 4-6. This also results in a noticeable difference in the number of human samples at these turning points, which are 14, 9, and 5 respectively. The primate samples that appear at the turning point in Figure 4, are p33, p24, p20, p27, and p15; those at the turning point, p2, p33, and p22; and those at the turning point in Figure 6, are p1, p26, p24, and p13.

Figure 6: The Similarity of SNP profiles among all samples in CGLG relative to modern human sample p6. Note: Levenshtein distance; Needleman_Wunsch. For detailed sample distribution along the X - aixs can be seen in Supplementary file 1-3.

Hu et al. used a coalescent model to predict past human population sizes from 3154 present-day human genomes [23]. The model detected a reduction in human ancestor population size from about 100,000 to 1280 individuals between around 930,000 and 813,000 years ago. The described bottleneck is congruent with a substantial chronological gap in the available African and Eurasian fossil record, and suggests coincident speciation event. It is interesting to tackle whether such an event was accompanied with the potential LG/CG/LGCG turning points described in this study. Therefore, it will be very valuable to continuously collect fossil DNA information from this period (930,000 and 813,000 years ago) or samples that have a direct genetic connection with this period in the future.

Conclusion

This study examined the genetic differences of 170 whole-genomes at 239 SNP loci of 18 Language Genes (LG) and 223 SNP loci of 18 Cognitive Genes (CG), clustered the SNP data of the above samples through PCA, and calculated the SNP pattern similarity between each sample under three perspectives: LG, CG, and CGLG. The basic conclusions include: (1) Both the polymorphic patterns of language genes and cognitive genes have undergone different evolutionary stages; (2) The polymorphic patterns of language genes and cognitive genes show significant differences in their early manifestations during human evolution, as reflected in the early patterns of SNP diversity in language genes being very similar across all animal samples, whereas cognitive genes exhibit significant variation right from the start among the selected animal samples; (3) It seems that samples from all five continents can be seen at every stage of evolution, indicating that new evolutionary populations have always had enough time to spread among the continents.

Acknowledgment

This study was supported by State Language Commission Research Grant (YB135-117), Association of Chinese Graduate Education Grant (B-2017Y0505-079) and National Research Center for Foreign Language Education Grant (ZGWYJYJJ10A042).

References

1. Hillert DG. On the evolving biology of language. Front Psychol. 2015;6:167436.

   [Crossref] [Google Scholar] [PubMed]

2. Stout D, Chaminade T. Stone tools, language and the brain in human evolution. Philos. Trans. R. Soc. B: Biol. Sci 2012;367(1585):75-87.

   [Crossref] [Google Scholar] [PubMed]

3. Tattersall I. An evolutionary framework for the acquisition of symbolic cognition by Homo sapiens. Comp. Cogn. Behav. Rev. 2008;3.

   [Crossref] [Google Scholar]

4. Fitch WT. The evolution of speech: A comparative review. Trends Cogn Sci. 2000;4(7):258-267.

   [Crossref] [Google Scholar] [PubMed]

5. Bolhuis JJ, Tattersall I, Chomsky N, Berwick RC. How could language have evolved?. PLoS Biol. 2014;12(8):e1001934.

   [Crossref] [Google Scholar] [PubMed]

6. Stout D. Stone toolmaking and the evolution of human culture and cognition. Philos. Trans. R. Soc. B: Biol. Sci. 2011;366(1567):1050-1059.

   [Crossref] [Google Scholar] [PubMed]

7. Prat S. Emergence of the genus Homo: From concept to taxonomy. L'anthropologie. 2022;126(4):103068.

   [Crossref] [Google Scholar]

8. Neubauer S, Hublin JJ, Gunz P. The evolution of modern human brain shape. Sci. Adv. 2018;4(1): eaao5961.

   [Crossref] [Google Scholar] [PubMed]

9. Li M, Zhang W, Zhou X. Identification of genes involved in the evolution of human intelligence through combination of inter-species and intra-species genetic variations. PeerJ. 2020;8:e8912.

   [Crossref] [Google Scholar] [PubMed]

10. Goriounova NA, Mansvelder HD. Genes, cells and brain areas of intelligence. Front. Hum. Neurosci. 2019; 13:390595.

    [Crossref] [Google Scholar] [PubMed]

11. Savage JE, Jansen PR, Stringer S, Watanabe K, Bryois J, De Leeuw CA, et al. Genome-wide association meta-analysis in 269,867 individuals identifies new genetic and functional links to intelligence. Nat. Genet. 2018;50(7):912-919.

    [Crossref] [Google Scholar] [PubMed]

12. Sniekers S, Stringer S, Watanabe K, Jansen PR, Coleman JR, Krapohl E, et al. Genome-wide association meta-analysis of 78,308 individuals identifies new loci and genes influencing human intelligence. Nat. Genet. 2017;49(7):1107-1112.

    [Google Scholar] [PubMed]

13. Xia W, Zhang Z. Language gene polymorphism pattern survey provided important information for education context in human evolution. biorxiv. 2022:10.

    [Crossref] [Google Scholar]

14. Shi L, Hu E, Wang Z, Liu J, Li J, Li M, et al. Regional selection of the brain size regulating gene CASC5 provides new insight into human brain evolution. Hum Genet. 2017;136(2):193-204.

    [Crossref] [Google Scholar] [PubMed]

15. Tattersall I. Endocranial volumes and human evolution. F1000Res. 2023;12.

    [Crossref] [Google Scholar] [PubMed]

16. Xia W, Zhang Z. Language gene polymorphism patterns: Important information on human evolution. J Data Mining Genomics Proteomics. 2023; 14:316.
17. Levenshtein VI. Binary codes capable of correcting deletions, insertions, and reversals. InSoviet Dokl. Phys 1966;10(8):707-710.

    [Google Scholar]

18. Smith TF, Waterman MS. Identification of common molecular subsequences. J. Mol. Biol. 1981;147(1):195-197.

    [Crossref] [Google Scholar] [PubMed]

19. Needleman SB, Wunsch CD. A general method applicable to the search for similarities in the amino acid sequence of two proteins. J. Mol. Biol. 1970;48(3):443-453.  

    [Crossref] [Google Scholar] [PubMed]

20. Yang MA, Gao X, Theunert C, Tong H, Aximu-Petri A, Nickel B, et al. 40,000-year-old individual from Asia provides insight into early population structure in Eurasia. Curr Biol. 2017;27(20):3202-3208.

    [Crossref] [Google Scholar] [PubMed]

21. Hublin JJ, Ben-Ncer A, Bailey SE, Freidline SE, Neubauer S, Skinner MM, et al. New fossils from Jebel Irhoud, Morocco and the pan-African origin of Homo sapiens. Nature. 2017;546(7657):289-292.

    [Crossref] [Google Scholar] [PubMed]

22. Richter D, Grün R, Joannes-Boyau R, Steele TE, Amani F, Rué M, et al. The age of the hominin fossils from Jebel Irhoud, Morocco, and the origins of the middle stone age. Nature. 2017;546(7657):293-296.

    [Crossref] [Google Scholar] [PubMed]

23. Hu W, Hao Z, Du P, Di Vincenzo F, Manzi G, Cui J, et al. Genomic inference of a severe human bottleneck during the early to middle pleistocene transition. Science. 2023;381(6661):979-984.

    [Crossref] [Google Scholar] [PubMed]