Study on Dynamics and Cluster Analysis between Carbon-based Nanoparticles and Amino Acids

We research the interaction between six representative carbon-based nanoparticles (CBNs) and 20 standard amino acids through all-atom molecular dynamics simulations. The six carbon-based nanoparticles are fullerene(C60), CNT55L3, CNT1010L3, CNT1515L3, CNT2020L3, and two-dimensional-graphene(Graphene33). Their curvatures decrease sequentially, and all of CNT are single-walled carbon nanotubes. We have observed that as the curvature of CBNs decreases, the adsorption effect of 20 amino acids with them has an increasing trend. In addition, we also used multi-dimensional clustering to analyze the adsorption effects of 20 amino acids on six carbon-based nanoparticles. We observed that the π-π interaction still plays an extremely important role in the adsorption of amino acids on carbon-based nanoparticles. Individual long-chain amino acids and “Benzene-like” Pro also have a strong adsorption effect with carbon-based nanoparticles.


Introduction
In recent years, the interaction between nanoparticles and proteins has attracted widespread attention [1; 2] . Experimental and theoretical simulation studies between nanoparticles and proteins [3] have been in continuous progress [4] , which is of great significance for the research of biopharmaceuticals, especially targeted drugs [5] . Various gold nanoparticles modified with different peptide chain groups have become important research objects for cancer-targeted drugs [6] . In addition, recent interactions between two-dimensional nanomaterials and individual nucleotides [7] indicate that this twodimensional screening structure of in-plane graphene and hexagonal boron nitride arrays may inspire the development of laboratory-on-a-chip sequencing in the future.
Carbon-based nanoparticles, as the most basic nanoparticles, have been widely concerned about the molecular dynamics research between them and proteins [8][9][10][11][12] . At present, many researchers have used molecular dynamics to study the interaction between carbon-based nanoparticles and proteins [5; 13-17] .
The interaction between amino acids and carbon-based nanoparticles or modified carbon-based nanoparticles has attracted wide attention from researchers in the experimental, quantitative calculation, and molecular dynamics [10; 12; 18] .
Experimentally, the interaction between carbon-based nanoparticles and amino acids, the unique electronic and optical properties of single-walled carbon nanotubes [23] , coupled with their size and mechanical strength, make single-walled carbon nanotubes become the key to the sensing platform [9] to the development of the next generation of biological transmission. Experimental studies have shown that carbon-based nanomaterials can promote neuronal electrical signals [16] . In addition, researchers have developed a variety of electrochemical biosensors based on carbon nanotubes [17] .
Moreover, in terms of drug delivery, carbon-based nanoparticles, especially those modified by peptides, have always received extensive attention from biomedical workers [24] ; in terms of quantitative calculations, the study of amino acids or amino acids in a vacuum computing environment The interaction between aromatics and carbon nanotubes [4; 10; 11] , and this part of the research content is also an important reference for studying the interaction of 20 amino acids and carbon-based nanoparticles in aqueous solutions; For all-atom molecular dynamics, on the one hand, some researchers have studied the interaction between some amino acids and a single carbon nanotube in one system [14] .
However, there is still a lack of molecular dynamics research between a single amino acid and a variety of carbon-based nanoparticles of different sizes in an aqueous environment, especially the corresponding research with a long simulation time. The comprehensive all-atom simulation study between carbon-based nanoparticles and 20 representative amino acids is very important: (1) This work is an important preliminary work for multi-scale research between carbon-based nanoparticles and ultra-large proteins; (2) The research is a paradigm for the study of interactions between other nanomaterials and 20 amino acids, and it is also the basis for future research on the relationship between complex nanomaterials and their interactions; (3) The work is also a study on the binding of nanoparticles and amino acid-related small molecules or proteins. It also as an important supporting material for prediction.
Our research has done a detailed studying on the interaction between a single amino acid and six carbon-based nanoparticles at the full-atom level. We tried to explore the relationship between 20 standard amino acids and 6 kinds of carbon-based nanoparticles from two views. For the view of carbon-based nanoparticles, under the premise of excluding the volume limitation, will nanoparticles with different curvatures affect the adsorption effect between them and amino acids; For the another view of 20 amino acids, different groups of amino acids are compared with 6 Will the adsorption effect of a representative carbon-based nanoparticle be affected by its type? Whether it is hydrophilic-hydrophobic interaction, π-π interaction, the positive/negative charged side chain, or simply the length of the side chain, these may affect the adsorption effect between a single amino acid and carbon-based nanoparticles. In addition, we hope that cluster analysis in machine learning can well characterize the adsorption strength between six representative carbon-based nanoparticles (CBNs) and 20 amino acids.
Next, let us explore the interactions between six representative carbon-based nanoparticles (CBNs) and 20 amino acids from the perspective of all-atom molecular dynamics simulations.

Simulation system
In our research, 120 molecular systems (6CBNs×20Aminoacids×3) were constructed, and the interaction between 6 carbon-based nanoparticles and 20 amino acids (Figure1BC) in an aqueous solution (Figure1A) was studied, and each study The system has done 3 repeated simulation studies, with a total of more than 360 simulation research trajectories, each trajectory 200ns, we try to objectively statistical research from the All-atom level. Among them, the six types of carbon-based nanoparticles are fullerene (C60), single-walled carbon nanotube (CNT55L3, CNT1010L3, CNT1515L3, CNT2020L3), and graphene (graphene33). In our research, the C60 is the carbon-based nanoparticle with the largest curvature (1/r), and the graphene33 with the smallest curvature is close to the planar state.  Table: the spatial size list of six representative carbon nanoparticles and amino acid systems From CNT55L3, CNT1010L3, CNT1515L3 to CNT2020L3, as shown in Figure   1A, the radii of carbon nanotubes with different radii increase, and the curvature decreases. The parameters of each carbon-based nanoparticle are shown in Figure1D- Table. In our research, to exclude the possibility of amino acids entering the carbon nanotubes as the CNT radius increases sequentially, we have used periodic boundary conditions. In other words, we studied the periodic infinitely long carbon tube CNT and the carbon-based plane Graphene, which allows us to focus only on the effect of the curvature of the carbon-based nanoparticle on its adsorption effect.
In addition, in order to eliminate the influence of CBNs volume on the adsorption effect, when constructing the system, we followed the principle of the same surface contact volume ratio of all simulation systems, namely

= −
The φ value of all simulation systems is fixed, that is, the surface entropy of the adsorption space of the system is a fixed value. Among them, represents the surface area of carbon-based nanoparticles, V represents the volume of the system, and represents the volume of carbon-based nanoparticles.
Our research, based on previous studies of molecular dynamics between carbonbased nanoparticles and proteins by other researchers, has grouped 20 amino acids. As shown in Figure  amino acids are shown in Figure 1C. Among them, the amino acids of the non-polar amino acid group start from amino acids 1-7, and the side chains increase sequentially.
The shortest side chain is 1-Gly, and the longest is 7-Met; the amino acids of the polar non-charged amino acid group are from 8-12 In the positively charged amino acid group, we put 18-His with a benzene ring at the end, but the 17-Arg in this group is the positively charged amino acid with the longest side chain.

Simulate system MD parameters
A molecular system of 120 simulation systems (6CBNs×20Aminoacids) in an aqueous solution, among which we selected the most commonly used SPC216 model for water molecules. Among them, the larger carbon-based nanoparticles are the CNT2020L3 system, and we have added 21,245 water molecules. We use Gromacs software [25; 26] throughout our molecular dynamics research. Our study used the frogleap method simulation, and the simulation integration step as 2fs. To ensure sufficient adsorption time between amino acids and CBNs, the duration of each simulation was 200ns. Ensure that our all-atom simulation time is sufficient to extract sufficient conformational information to prepare for the subsequent multi-scale coarse-grained force field. In the initial stage, the distance between amino acids and CBNs is about 1.5 nm. Our simulation systems are all NVT systems [27] . The simulated temperature of the system is 330K, and we use the Berendsen temperature adjustment method. We used periodic boundary conditions to make our carbon nanotubes periodic infinite length and infinite graphene plane [26] . This paper mainly studies the adsorption trend between 6 representative carbonbased nanoparticles and 20 standard amino acids, as shown in Figure 2A. Our research found that as the curvature decreases, the adsorption rate shows an increasing trend, as shown in Figure 2B. Moreover, our research also makes suitable use of multidimensional clustering analysis to describe the overall change in its adsorption rate, when 20 amino acids change with the decreasing curvature of carbon-based nanoparticles. As shown in Figure 2C, the purple line is the short-chain amino acid 01-Gly, the yellow line is the hydrophobic amino acid 07-Met with the longest side chain, and the green line is the aromatic amino acid 15-Trp with the longest side chain. We can observe that the adsorption strength of the three amino acids is: 15-Trp>07-Met>01-Gly. In other words, the adsorption strength of aromatic amino acids is much greater than that of amino acids with the longest side chain.

Kinetics Analysis
In our work, the minimum contact distance Mindist changes with time for each group of amino acids. The change of the minimum contact distance with time is a very important indicator. In many studies of protein and nanoparticles, this indicator is used to describe the change of the distance from the surface of the nanoparticle with time.
The reason is that different carbon-based nanoparticles have different radii, and choosing the Mindist indicator can better reflect the adsorption between amino acids and carbon-based nanoparticles. Next, we will discuss the kinetic characteristics of different groups of amino acids Mindist over time.

Figure3
The change of the minimum distance between CBNs and amino acids over time, the horizontal axis t (ns), the vertical axis Mindist (nm). From C1-C6 respectively: Ⅰ group picture C1 orange line C60, group Ⅱ picture C2 yellow line CNT55L3, group Ⅲ picture C3 bright green line CNT1010L3, group Ⅳ picture C4 green line CNT1515L3, group Ⅴ picture C5 bright blue line CNT2020L3, Ⅵ Group picture C6 purple line  We will analyze the statistical distribution from the data that the minimum distance changes over time. This research step is very necessary because, in Mindist data, some data smaller than 0.3 nm are actually in dynamic changes. As shown in Figure4, we can macroscopically observe the adsorption relationship between each group of amino acids and CBNs. Among them, the adsorption effect of aromatic amino acids and CBNs is the best, and their First-peak can be greater than 0.225, while the first-peak of negatively charged amino acids hovers around 0.1, and the overall is less than 0.15.

Statistical Analysis
Non-polar amino acids have different adsorption strengths from CBNs according to their side chains and chain lengths; polar non-charged amino acids have different internals, but the overall adsorption strength is lower than non-polar amino acids; The strongest positively charged amino acids is comparable to aromatic amino acids; But the overall adsorption strength of negatively charged amino acids is weak.

(b)The adsorption between CBNs and 20 amino acids
(1) The average adsorption strength. Figure 2B shows the average adsorption strength of 20 amino acids for 6 kinds of CBNs. We will calculate the First-peak minimum distance distribution of the CBNs obtained from Figure 4 on the 20 amino acids, that is, the amino acids corresponding to the same CBNs are added and averaged.
As a result, we are pleased to observe from Figure 2B that as the curvature of CBNs becomes smaller, the adsorption strength of carbon-based nanoparticles for 20 amino acids has an obvious upward trend. This also shows that Average-mindist-distribution-First-peak can well describe the influence of CBNs curvature changes on the adsorption strength of 20 amino acids. In other words, the surface curvature of carbon-based nanoparticles has an obvious effect on the adsorption between them and amino acids. (2) The adsorption strength with 0.3 nm. In Figure 5, we tried to count the changes in the overall adsorption strength of different amino acids when the minimum distance between CBNs and 20 amino acids was less than 0.3nm. We ranked each group of amino acids. Among them, we set different color settings for the adsorption statistics of different carbon-based nanoparticles. We can see that, as shown in Figure 5 nanoparticles. However, it is worth mentioning that 3-Pro, a "benzene-like" amino acid, shows excellent adsorption capacity with CBNs.
(c)The multi-dimensional cluster analysis We hope to characterize the influence of 20 amino acids on the curvature changes of CBNs through machine learning multi-dimensional cluster analysis. The multidimensional cluster analysis is shown in Figure 6, the horizontal axis represents the curvature of CBNs, and the vertical axis represents the Mindist-distribution-first-peak of amino acids. (CBNs) and 20 amino acids through all-atom molecular dynamics simulations.
Constructed 120 molecular systems (6CBNs×20Aminoacids), studied the interactions between 6 carbon-based nanoparticles and 20 amino acids in aqueous solutions, and performed 3 repeated simulation studies for each research system, totaling more than 360 Simulate research trajectories, each trajectory 200ns, trying to objectively statistical research from the all-atom level.
We have observed that as the curvature of carbon-based nanoparticles decreases, the adsorption effect of 20 amino acids with them is getting better and better. In addition, we also used multi-dimensional clustering to analyze the adsorption effects of 20 amino acids on six carbon-based nanoparticles. We observed that the π-π interaction plays an extremely important role in the adsorption of amino acids on carbon-based nanoparticles. Individual long-chain amino acids also have a strong adsorption effect with carbon-based nanoparticles. Specifically, the special hydrophobic amino acid 03-Pro, the long-chain aromatic amino acid 15-Trp, and the positively charged amino acid 17-Arg, these three amino acids are far ahead in the adsorption capacity of carbon-based nanoparticles; the remaining two aromatic amino acids The adsorption effect of 13-Phe and 14-Tyr is better than that of other amino acids. The regular adsorption effect between CBNs and amino acids can also bring certain inspiration to the new generation of nano-biocomputers.
In addition, this research work as a full-atom simulation of carbon-based nanoparticles and 20 kinds of amino acids has laid a solid simulation data foundation for further multi-scale coarse-grained force field extraction.