Protein Evolution

How Accurate is the Technique?

Molecular knowledge = knowledge per site x number of sites.

Each site is a column in a multiple sequence alignment. So insulin has 52 sites ( figure 4.3). The knowledge per site is as follows:

Knowledge per site = log₂ ( 64 possible condons/ allowed codons).

The big uncertainty in this definition is which codons are allowed. Allowed codons are codons that result in amino acids that yield either a fully functional or marginally functional protein. In other words, an allowed amino acid does not completely destroy the protein’s function. If the number of allowed codons can be calculated accurately at each site, then the molecular knowledge so calculated can be related to a probability of evolution. The equation is as follows:

Odds of a protein evolving = 1 in 2 ^{(molecular knowledge)} chance.

        The error in this technique arises because it is not easy to determine which amino acids are allowed. The number of allowed amino acids is not the number of amino acids found in a particular column in a multiple sequence alignment. Natural selection tends to weed out functional proteins that are slightly deleterious. This means that in many columns the number of amino acids that are allowed is significantly greater than what is observed in nature.

        There are two ways to determine the allowed amino acids. One is through experimentation. The other is to develop a set of rules that allow additional amino acids at each site in a protein. This chapter proposed a set of rules that grouped amino acids into 9 groups and proposed a method to determine the true number of allowed amino acids at each site. Since these rules are arbitrary, the most pressing question is do these rules accurately predict allowed amino acids?

        At first glance, the rules seem too stringent. For example, at position 24 in table 4.5 only valine is found. With these rules, leucine, isoleucine, methionine and alanine are also allowed because they belong to the same amino acid group. Experimental evidence shows that if this valine is replaced with leucine, the resulting insulin will lose more than 90% of its functionality.² Replacing it with an amino acid from a different group should further destroy functionality (leaving the insulin molecule with almost no functionality). Positions 25 and 26 of the B chain are always tyrosine or phenylalanine. Replacing position 26 with a serine almost completely destroys insulin functionality.¹ Position 25 is not as critical because a serine here only destroys 85% of insulin’s functionality.¹ A single amino acid substitution can in many cases result in a non-functional protein.

        While no analytical technique designed to calculate molecular knowledge will be perfect, the one introduced in this chapter in most cases should be close to the true value of molecular knowledge. The only way to validate this analytical technique is to compare its predictions to experimental evidence. Researchers can directly substitute amino acids at each site in a protein and then screen the resulting protein for functionality. From this experimental evidence, it is possible to directly calculate molecular knowledge.

Appendix 6 compares the analytical techniques developed in this chapter to experimental results, and these comparisons support figure 4.9. In this figure, the actual knowledge is calculated from the experimental techniques described in appendix 6. The knowledge predicted by the analytical technique agrees very well with the experimental data, but the two calculations are seldom identical. Sometimes the analytical technique yields more knowledge and sometimes it yields less. The shaded region in figure 4.9 represents this uncertainty. The agreement between the two approaches could be improved with a more complex set of rules to better estimate the allowed amino acids. Nevertheless, even with the simple set of rules proposed in this chapter, the calculations agree remarkably well.

Figure 4.9: Actual knowledge vs. Assigned Knowledge

Figure 4.10 on the next page should be compared to figure 4.1 at the beginning of this chapter. In figure 4.1, the technique used to calculate the information of insulin (as it exists today) is shown. This information cannot be related to a probability because insulin has already been optimized by natural selection. In figure 4.1, the only door is the last door. All of the doors leading up to this door are hidden. The technique introduced in this chapter attempts to reconstruct the combination of the earlier doors. In particular, the first door is important because it is this door that determines whether or not naturalistic laws can explain the evolution of insulin. The screen in figure 4.10 shows the combinations that will open the first door. For example, the first position is represented by an asterisk because all 20 amino acids are allowed at this position. In the second position, only gln is found today, but since asn belongs to the same group, asn is shown in parenthesis. Any combination with either gln or asn at the second position will open the door. The accuracy of this technique depends on how well it predicts the combination of the first door. Insulin may not imply design for reasons already discussed. Nevertheless, when other proteins are analyzed with this method, the design inference is very strong.

Figure 4.10: Molecular Knowledge of Insulin

Experimental Evidence for figure 4.9: In appendix 6, the techniques developed in this chapter are compared to experiments in which researchers randomly mutate proteins and then screen for functionality. Reading appendix 6, immediately after finishing this chapter is encouraged because the two subjects complement each other well.

Experimental Evidence for Figure 4.9 can be found in the following two pdf files.

appendix 6: experimental evidence from AAG

appendix 7: more experimental evidence from LacI
(appendix 7 is not in the book because of page constraints).

Home: Origin of Life and Intelligent Design