Derivatives of logarithmic stationary distributions for policy gradient reinforcement learning.

Most conventional policy gradient reinforcement learning (PGRL) algorithms neglect (or do not explicitly make use of) a term in the average reward gradient with respect to the policy parameter. That term involves the derivative of the stationary state distribution that corresponds to the sensitivity of its distribution to changes in the policy parameter. Although the bias introduced by this omission can be reduced by setting the forgetting rate gamma for the value functions close to 1, these algorithms do not permit gamma to be set exactly at gamma = 1. In this article, we propose a method for estimating the log stationary state distribution derivative (LSD) as a useful form of the derivative of the stationary state distribution through backward Markov chain formulation and a temporal difference learning framework. A new policy gradient (PG) framework with an LSD is also proposed, in which the average reward gradient can be estimated by setting gamma = 0, so it becomes unnecessary to learn the value functions. We also test the performance of the proposed algorithms using simple benchmark tasks and show that these can improve the performances of existing PG methods.

Identification of macrolide antibiotic-binding Human_p8 protein.

Clarithromycin is a macrolide antibiotic that is widely used in clinical medicine. Macrolide antibiotics such as clarithromycin specifically bind to the 50S subunit of the bacterial ribosome thereby interfering with protein biosynthesis. A selected peptide sequence from our former study, composed of 19 amino acids, which was isolated from a phage display library because of its ability to bind clarithromycin, displayed significant similarity to a portion of the human_p8 protein. The recombinant p8 protein binds to biotinylated-clarithromycin immobilized on a streptavidin-coated sensor chip and the dissociation constant was determined. The binding of recombinant p8 protein to double-stranded DNA was inhibited by biotinylated-clarithromycin, clarithromycin, erythromycin and azithromycin in gel mobility shift assay. Dechlorogriseofulvin, obtained from a natural product screening, also inhibited human p8 protein binding to DNA. This study illustrates the general utility of the phage display method in detecting protein-ligand interactions.

Identification of antibiotic clarithromycin binding peptide displayed by T7 phage particles.

Peptide libraries displayed by T7 phage, which contain random cDNA fragments insets, were screened for their ability to bind to a biotinylated derivative of clarithromycin. Phage particles bound to an immobilized derivative of the antibiotic were isolated and the inserted cDNA was amplified and sequenced. A common selected peptide sequence, composed of 19 amino acids, was obtained and a synthetic peptide with this sequence was produced. Surface plasmon resonance experiments showed that the synthetic peptide immobilized on a sensor chip bound to clarithromycin and the dissociation constant was determined to be 2.1 x 10(-3) M. The dissociation constants of other macrolide antibiotics, erythromycin, roxithromycin, azithromycin and josamycin were also determined to be 5.4 x 10(-3) M, 6.2 x 10(-5) M, 1.1 M and 3.4 x 10(-2) M, respectively. These results indicated that T7 phage display method might be useful to determine relatively weak interactions between small molecule drugs and the selected peptides which could represent a possible binding site conserved in binding proteins.