Prediction of Protein pKa with Representation Learning

Author(s):  
Hatice Gokcan ◽  
Olexandr Isayev

The behavior of proteins is closely related to the protonation states of the residues. Therefore, prediction and measurement of pKa are essential to understand the basic functions of proteins. In this work, we develop a new empirical scheme for protein pKa prediction that is based on deep representation learning. It combines machine learning with atomic environment vector (AEV) and learned quantum mechanical representation from ANI-2x neural network potential (J. Chem. Theory Comput. 2020, 16, 4192). The scheme requires only the coordinate information of a protein as the input and separately estimates the pKa for all five titratable amino acid types. The accuracy of the approach was analyzed with both cross-validation and an external test set of proteins. Obtained results were compared with the widely used empirical approach PROPKA. The new empirical model provides accuracy with MAEs below 0.5 for all amino acid types. It surpasses the accuracy of PROPKA and performs significantly better than the null model. Our model is also sensitive to the local conformational changes and molecular interactions.

1981 ◽  
Vol 46 (3) ◽  
pp. 772-780 ◽  
Author(s):  
Jorga Smolíková ◽  
Jan Pospíšek ◽  
Karel Bláha

Infrared spectra of the L-alanine (I), L-leucine (II), L-valine (III) and L-tert-leucine (IV) N-acetyl N'-methylamides were measured. Amides I-IV are not self associated in tetrachlormethane in the concentration 2 . 10-5 mol l-1 at room temperature and in tetrachloroethylene in the concentration 1.5 . 10-4 mol l-1 at temperatures above 65° C. True conformational changes are observable only with the least flexible amide IV which exists at room temperature in a C5 conformation. This conformational type is also highly populated in the valine derivative III, but is less important in the alanine and leucine derivatives I and II in which the intramolecularly bonded C7 and the distorted hydrogen-nonbonded conformations contribute seriously.


Author(s):  
Karan Aggarwal ◽  
Shafiq Joty ◽  
Luis Fernandez-Luque ◽  
Jaideep Srivastava

Sufficient physical activity and restful sleep play a major role in the prevention and cure of many chronic conditions. Being able to proactively screen and monitor such chronic conditions would be a big step forward for overall health. The rapid increase in the popularity of wearable devices pro-vides a significant new source, making it possible to track the user’s lifestyle real-time. In this paper, we propose a novel unsupervised representation learning technique called activ-ity2vecthat learns and “summarizes” the discrete-valued ac-tivity time-series. It learns the representations with three com-ponents: (i) the co-occurrence and magnitude of the activ-ity levels in a time-segment, (ii) neighboring context of the time-segment, and (iii) promoting subject-invariance with ad-versarial training. We evaluate our method on four disorder prediction tasks using linear classifiers. Empirical evaluation demonstrates that our proposed method scales and performs better than many strong baselines. The adversarial regime helps improve the generalizability of our representations by promoting subject invariant features. We also show that using the representations at the level of a day works the best since human activity is structured in terms of daily routines.


2010 ◽  
Vol 37-38 ◽  
pp. 9-13
Author(s):  
Hong Xin Wang ◽  
Ning Dai

A non-iterative design method about high order intermittent mechanisms is presented. The mathematical principle is that a compound function produced by two basic functions, and then one to three order derivatives of the compound function are all zeroes when one order derivative of each basic function is zero at the same moment. The design method is that a combined mechanism is constructed by six bars; the displacement functions of the front four-bar and back four-bar mechanisms are separately built, let one order derivatives of two displacement functions separately be zero at the same moment, and then get geometrical relationships and solution on the intermittent mechanism. A design example shows that this method is simpler and transmission characteristics are better than optimization method.


2021 ◽  
pp. 1-8
Author(s):  
Adeyeye EI ◽  
◽  
Idowu OT ◽  

This article reports the amino acid composition of the Nigerian local cheese called ‘wara’. ‘Wara’ is made by boiling cow milk with some added coagulant to cuddle the milk protein resulting in coagulated milk protein and whey. ‘Wara’ used to be an excellent source of nutrients such as proteins, fats, minerals and vitamins. Samples were purchased in Ado-Ekiti, Nigeria. Amino acid values were high (g/100g crude protein) in Leu, Asp, Glu, Pro, Phe, Arg with total value of 97.7. The quality parameters of the amino acids were: TEAA (42.6g/100g and 43.6%) whereas TNEAA (55.1g/100g and 56.4%); TArAA (12.8g/100g and 13.1%); TBAA (14.2g/100g and 14.5%); TSAA (3.10g/100g and 3.17%); %Cys in TSAA (51.4); Leu/Ile ratio (1.74); P-PER1 (2.65); P-PER2 (2.48); P-PER3 (2.41); EAAI1 (soybean standard) (1.29) and EAAI2 (egg standard) (99.9); BV (97.2) and Lys/Trp ratio (3.62). The statistical analysis of TEAA/TNEAA at r=0.01 was not significantly different. On the amino acid scores, Met was limiting (0.459) at egg comparison, Lys was limiting at both FAO/WHO [24] and preschool EAA requirements with respective values of 0.966 and 0.97. Estimates of essential amino acid requirements at ages 10-12 years (mg/kg/day) showed the ‘wara’ sample to be better than the standard by 3.72-330% with Lys (3.72%) being least better and Trp (330%) being most. The results showed that ‘wara’ is protein-condensed which can be eaten as raw cheese, flavoured snack, sandwich filling or fried cake.


2021 ◽  
Author(s):  
Ahmad Shahir Sadr ◽  
zahra abdollahpour ◽  
Atousa Aliahmadi ◽  
Changiz Eslahchi ◽  
Mina Nekouei ◽  
...  

Abstract The hydrogen/deuterium exchange (HDX) is a reliable method to survey the dynamic behavior of proteins and epitope mapping. Matrix-Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry (MALDI-TOF MS) is a quantifying tool to assay for HDX in the protein of interest. We combined HDX-MALDI-TOF MS and molecular docking/MD simulation to identify accessible amino acids and analyze their contribution in the structural changes of profilin1 (PFN1). The molecular docking/MD simulations are computational tools for enabling the analysis of the type of amino acids that may be involved via HDX identified under the lowest binding energy condition. Glycine to Valine amino acid (G117V) substitution mutation is linked to amyotrophic lateral sclerosis (ALS). This mutation is found to be in the actin-binding site of PFN1 and prevents the dimerization/polymerization of actin and invokes a pathologic toxicity that leads to ALS. In this study, we sought to understand the PFN1 protein dynamic behavior using purified wild type and mutant PFN1 proteins. The data obtained from HDX-MALDI-TOF MS for PFN1WT and PFN1G117V at various time intervals, from seconds to hours, revealed multiple peaks corresponding to molecular weights from monomers to multimers. PFN1/Benzaldehyde complexes identified 20 accessible amino acids to HDX that participate in the docking simulation in the surface of WT and mutant PFN1. Consistent results from HDX-MALDI-TOF MS and docking simulation predict candidate amino acid(s) involved in the dimerization/polymerization of PFNG117V. This information may shed critical light on the structural and conformational changes with details of amino acid epitopes for mutant PFN1s’ dimerization, oligomerization, and aggregation.


eLife ◽  
2014 ◽  
Vol 3 ◽  
Author(s):  
Bargavi Thyagarajan ◽  
Jesse D Bloom

Influenza is notable for its evolutionary capacity to escape immunity targeting the viral hemagglutinin. We used deep mutational scanning to examine the extent to which a high inherent mutational tolerance contributes to this antigenic evolvability. We created mutant viruses that incorporate most of the ≈104 amino-acid mutations to hemagglutinin from A/WSN/1933 (H1N1) influenza. After passaging these viruses in tissue culture to select for functional variants, we used deep sequencing to quantify mutation frequencies before and after selection. These data enable us to infer the preference for each amino acid at each site in hemagglutinin. These inferences are consistent with existing knowledge about the protein's structure and function, and can be used to create a model that describes hemagglutinin's evolution far better than existing phylogenetic models. We show that hemagglutinin has a high inherent tolerance for mutations at antigenic sites, suggesting that this is one factor contributing to influenza's antigenic evolution.


2021 ◽  
Author(s):  
Sarah E Leininger ◽  
Judith Rodriguez ◽  
Quyen V Vu ◽  
Yang Jiang ◽  
Ma Suan Li ◽  
...  

The speed of protein synthesis can dramatically change when consecutively charged residues are incorporated into an elongating nascent protein by the ribosome. The molecular origins of this class of allosteric coupling remain unknown. We demonstrate, using multi-scale simulations, that positively charged residues generate large forces that pull the P-site amino acid away from the A-site amino acid. Negatively charged residues generate forces of similar magnitude but opposite direction. And that these conformational changes, respectively, raise and lower the transition state barrier height to peptide bond formation, explaining how charged residues mechanochemically alter translation speed. This mechanochemical mechanism is consistent with in vivo ribosome profiling data exhibiting a proportionality between translation speed and the number of charged residues, experimental data characterizing nascent chain conformations, and a previously published cryo-EM structure of a ribosome-nascent chain complex containing consecutive lysines. These results expand the role of mechanochemistry in translation, and provide a framework for interpreting experimental results on translation speed.


2005 ◽  
Vol 187 (10) ◽  
pp. 3502-3510 ◽  
Author(s):  
Shintaro Seto ◽  
Atsuko Uenoyama ◽  
Makoto Miyata

ABSTRACT Several mycoplasma species are known to glide on solid surfaces such as glass in the direction of the membrane protrusion, but the mechanism underlying this movement is unknown. To identify a novel protein involved in gliding, we raised monoclonal antibodies against a detergent-insoluble protein fraction of Mycoplasma mobile, the fastest glider, and screened the antibodies for inhibitory effects on gliding. Five monoclonal antibodies stopped the movement of gliding mycoplasmas, keeping them on the glass surface, and all of them recognized a large protein in immunoblotting. This protein, named Gli521, is composed of 4,738 amino acids, has a predicted molecular mass of 520,559 Da, and is coded downstream of a gene for another gliding protein, Gli349, which is known to be responsible for glass binding during gliding. Edman degradation analysis indicated that the N-terminal region is processed at the peptide bond between the amino acid residues at positions 43 and 44. Analysis of gliding mutants isolated previously revealed that the Gli521 protein is missing in a nonbinding mutant, m9, where the gli521 gene is truncated by a nonsense mutation at the codon for the amino acid at position 1170. Immunofluorescence and immunoelectron microscopy indicated that Gli521 localizes all around the base of the membrane protrusion, at the “neck,” as previously observed for Gli349. Analysis of the inhibitory effects of the anti-Gli521 antibody on gliding motility revealed that this protein is responsible for force generation or force transmission, a role distinct from that of Gli349, and also suggested conformational changes of Gli349 and Gli521 during gliding.


Author(s):  
M.F. Fuller ◽  
R. McWilliam ◽  
T.C. Wang

The optimal balance of amino acids in the diet of the growing pig was estimated by ARC (1981) on the basis of a number of disparate studies augmented by data on the amino acid composition of the whole body on the premise that the amino acids incorporated into accreted body proteins are the major determinant of requirements and that this pattern is not distorted by inequalities in the utilisation of individual amino acids. In an accompanying paper (Wang & Fuller, paper no. 91) an optimal pattern was derived by direct experiment which was shown to be utilised better than that described by ARC (1981). That pattern, however, which related to one particular rate of nitrogen input and the particular rate of protein accretion which that input supported, includes two components, a requirement for maintenance and a requirement for protein accretion. There is clear evidence from studies with rats and chicks that the optimal pattern of amino acids for maintenance and growth are quite different and so the optimal pattern for any particular rate of growth will depend on the relative contributions of the two components. The purpose of this experiment was to estimate both.


Sign in / Sign up

Export Citation Format

Share Document