Exam II Flashcards

Question

drug hit VS drug lead

Answer 1

Drug hit: a molecule found in a HTS that has activity Optimized hits → leads Drug lead: hits that are chemically “tweaked” (optimized) to improve selectivity, potency, or pharmacokinetics (how compound is metabolized)

Answer 2

any computational technique that helps identify potential drugs, without having to make and test them in reality accelerate drug discovery without the time and expense of HTS Goal is to prioritize compounds for subsequent testing (so fewer need be tested costly and time-intensive

Answer 3

Cost Invested: $350 million per single drug; $5 billion per successful new medicine Time Invested: 11-14 years

Answer 4

CADD success is driven by advances in data, technology, and algorithms Becoming increasingly successful in the last 20 years due to: 1. Availability of structural information 2. Increases in computer hardware performance 3. New theories, algorithms, and software 4. Advances in computing power enabling larger molecular simulations

Answer 5

- critical for protein-ligand binding - measures disorder and probability of binding - how can system be arranged with same overall energy? -- molecule flexibility about single bonds (poses) *** systems tend to move to a higher entropy -- bc energy is required to go against entropy, which is bad for binding

Answer 6

Chains: more flexible so more entropically favorable Rings: more rigid, less entropic penalty bc change in entropy is not as much

Answer 7

1. Rigid “lock-and-key” -- Receptor and ligand fit perfectly together and are rigid 2. Induced FIt -- Induces a change in the shape of the binding pocket, so that the ligand can fit when bound *** More accurate theory (accounts for protein flexibility)

Answer 8

1. Receptor exists in a population of energetically low-lying sub-states (ensemble) 2. Ligand binds to one of these sub-states and shifts the population toward the favorite bound conformation --- Stabilizes energy when fit is compatible ** Population/conformational shift into different ensembles

Answer 9

1. Kd (Dissociation Constant) 2. Residence time 3. IC 50 & EC 50

Answer 10

Measures drug-receptor binding affinity Lower Kd → higher affinity Kd = Koff / Kon Fast Kon + slow Koff → good affinity Slow Kon + fast Koff → bad affinity Units: M-1s-1

Answer 11

Koff (Association Rate Constant) = how quickly does the ligand leave pocket/unbind Kon (Dissociation Rate Constant) = how quickly does the ligand enter pocket/bind Fast Kon + slow Koff → good affinity Slow Kon + fast Koff → bad affinity

Answer 12

Physical limits on how fast Kon can be At max speed, Kon is constrained by the rate molecules diffuse through water to target Many binding events involve induced fit, which can slow Kon below its theoretical diffusion limit

Answer 13

Residence time (T) = 1 / Koff Longer residence time → prolonged drug action Drugs with low Koff often have long-lasting effects even if plasma levels drop Covalent inhibitors effectively have near-infinite residence times

Answer 14

IC50 (Half-Maximal Inhibitory Concentration): Drug concentration needed to inhibit by 50%. EC50 (Half-Maximal Effective Concentration): Drug concentration needed for 50% max biological effect

Answer 15

Kd: true thermodynamic constant (intrinsic property of a ligand-protein interaction) ***Independent of receptor concentration IC50/EC50: depend on experimental set up ***Dependent of receptor concentration

Answer 16

orthosteric vs allosteric pockets

Answer 17

Where natural ligand binds, triggers protein function (primary or active site) Easier to design bc of defined binding pocket target

Answer 18

Other pocket separate from orthosteric site Ligands regulate activity (enhance or inhibit) w/o directly competing with natural ligand Disable on a range (minimizes not eliminates → helps with side effects)

Answer 19

- direct competition with endogenous ligands - more likely to fully block activity - known pockets - easy to ID - lacks selectivity - dose-limiting toxicity: more likely to fully inhibit drug targets, so could cause excessive responses - competition with endogenous ligands: drugs must be in high concentrations to outcompete natural ligands

Answer 20

- more selective targeting - fine-tuned control, partially inhibit/activate - encourages new drug designs, can target proteins that are difficult to affect with traditional (orthosteric) drugs - more diverse - easier to create allosteric agonists - harder to discover (sites are often hidden/cryptic) - may have weaker effects; need proteins natural ligand to be present to work effectively

Answer 21

SMILES: Simplified Molecular Input Line Entry System simplifies chemical structure representation into simple text string SMILE strings are NOT unique; many possible strings for a given molecule

Answer 22

Atoms: B, C, N, O, P, S, F, Cl, Br, or I - Aromatic atoms are LOWERCASE Bonds: Assumed to be single, unless otherwise specified - Double: = - Triple: # Branches: parentheses - CCC(C)CC Rings: carbons marked with numbers connect together with coordinating number - C1CCCCC1

Answer 23

ensures consistency in molecule representation Bc molecules can have multiple valid SMILES string

Answer 24

Advantages: easier to compress for database storage Disadvantages: SMILES only includes info about atoms/bonds, NOT atomic positions in 3D space

Answer 25

Encode both connectivity and limited spatial data (makes up for disadvantage of SMILES) Projected onto 2 coordinates Used for visualizing molecules in an easier way than 3D - # of atoms - # of bonds - bond type

Answer 26

Provide full molecular coordinates Used for storing small molecules

Answer 27

Describe biomolecular (e.g. protein) structures Do not typically need to specify the bonds for proteins bc every amino acid will have the same bonds every time Easier file to manipulate than 3D sdf - atom index - atom name - residue name - chain - residue index - XYZ Chain > residue > atom

Answer 28

The 1 (fifth field) suggests 3D coordinates. If this field is 0, it typically means 2D.

Answer 29

Powerful cheminformatics library for Python Useful for loading/viewing molecules

Answer 30

database of chemical molecules and their activities against biological assays. (NIH) Integrates chemical and biological data Lists contain experiments where drug was both active/inactive Supports versatile chemical information searches

Answer 31

Compounds searching: - By name - By substructure (substructure found in molecule matches sub of target molecule) - By similarity (Tanimoto) - By identity -Information available (SMILES, chemical properties) Bioassay searching: - Filtering - Information available

Answer 32

Loading a compound by name (e.g., aspirin) Retrieving: - Name of an existing compound - Properties of an existing compound - Bioassay activities of an existing compound Building a library of similar compounds in terms of: - Substructure - Superstructure - Tanimoto similarity

Answer 33

Bridges chemical and biological drug data “The DrugBank database is a unique bioinformatics and cheminformatics resource that combines detailed drug data with comprehensive drug target information.” Common Drugbank Tasks: 1. Chemical structure searching 2. Drug name searching 3. Target searching

Answer 34

Chemical structure searching Drug name searching Target searching

Answer 35

Substructure – Finds molecules that contain the query structure as part of their larger structure. Superstructure – Finds molecules that expand upon the query structure by adding more atoms or groups. Similarity – Finds molecules that resemble the query structure based on overall shape or properties.

Answer 36

1. Convert molecular substructures into sets - “Binary fingerprints descriptor” - 1 = substructure present, 0 = not present 2. Calculate Tanimoto (or Jaccard) index between the sets (finds similarity between 2 molecules) - Allows us to compare two sets of things → -- Size of intersection divided by size of union. --- Intersection: Number of bits in common --- Union: Number of bits in one or the other

Answer 37

finds similarity between 2 molecules Size of intersection divided by size of union. -- Intersection: Number of bits in common -- Union: Number of bits in one or the other c / a + b - c a = # of 1 bits in set A b = # of 1 bits in set B c = # of same 1 bits in both set A and B

Answer 38

Purpose: high-throughput or virtual screening to find potential drug candidates Well designed molecular libraries balance diversity and similarity

Answer 39

Goal is: 1. Structurally diverse molecules 2. Molecules that are similar to known hits 3. Molecules that are easy to synthesize

Answer 40

- use algorithm to get set of unique molecular representatives 1. For each molecule in your library, calculate the set of associated “nearest neighbors.” - Near here means sufficiently similar (Tanimoto coefficient). - You must pick the cutoff that defines “sufficiently similar.” 2. Are there any molecules that have no near neighbors? - These are “singletons.” - Remove them from the pool of molecules, but remember them as “centroid.” - Remove most popular centroid 3. Which compound has the largest set of nearest neighbors? - Remove those from the pool, but remember that particularly popular compound (a “centroid”). 4. Repeat step 3 until there are no remaining compounds in the pool. ***The set of centroids is your diversity set.

Answer 41

Measure similarity with tanimoto similarity Congeneric series aid in systematic drug optimization by using substructures or tanimoto similarity

Answer 42

Combinational libraries expand chemical diversity for drug discovery look at synthesizing pathways

Answer 43

Diverse molecular library when aiming to explore a broad chemical space and identify novel leads for a wide range of targets Similarity library of compounds is preferred for focused screening and structure-activity relationship (SAR) studies

Answer 44

Filtering by chemical properties improves drug-discovery efficiency Criteria for orally available drugs ***Allowed to violate 1 rule Lipinski's Rule of Fives: <= 5 nitrogens/oxygens with attached hydrogens (“hydrogen-bond donors”) <= 10 nitrogens/oxygens total (“hydrogen-bond acceptors”) Molecular mass < 500 daltons Octanol-water partition coefficient (logP) <= 5

Answer 45

Lipinski's Rule of Fives: <= 5 nitrogens/oxygens with attached hydrogens (“hydrogen-bond donors”) <= 10 nitrogens/oxygens total (“hydrogen-bond acceptors”) Molecular mass < 500 daltons Octanol-water partition coefficient (logP) <= 5 [sol in octanol] / [sol in H2O]

Answer 46

Need to filter bc there are SOOO many drugs out there(10^60), and we need to figure out which ones would be more likely to work within the library Resources can be limited (time & money)

Answer 47

Provides comprehensive chemical property data 1. Search for the compound 2. Chemical properties: molecular weight, LogP (XLogP3-AA), etc. 3. Other useful information (SMILES, known protein targets)

Answer 48

Filters improve screening library reliability Molecules with certain substructures often appear as hits in high-throughput screens, against many proteins. (false positives that are not worth optimizing)

Answer 49

Caused by Yersinia pestis Evidence of human infection for at least the past 6,000 years First plague (6th century): 50 million people About 200 million people on earth at the time Second plague (14th century): ⅓ of Europe Third plague (19th century): 12 million deaths in India and China Since 1965…dropped death rates for diseases using pharmaceuticals (antibiotics)

Answer 50

Bad scientist Everyone thought ulcers were caused by stress and spicy foods Dr. Marshall and Dr. John Warren thought it was H.pylori Bad experimental design… -- drink H. pylori, vomiting, so give antibiotics

Answer 51

Controversy: Costs a lot, but also abuses CEO/founder Turing Pharmaceuticals Got rights to the antiparasitic drug Daraprim, increased price from $13.50 to $750 overnight Convicted of securities fraud and conspiracy → prison Asked court for release in 2020 so he could develop a cure for COVID-19. Denied!!

Answer 52

1. Identify diseases 2. Isolate protein involved in disease -- 2-5 years 3. Find drug effective against disease protein -- 2-5 years 4. Preclinical testing -- 1-3 years 5. Formulation and scale-up 6. Human clinical trials -- 2-10 years 7. FDA approval -- 2-3 years

Answer 53

From luck From nature From medicinal chemistry

Answer 54

Penicillin (1928): Accidental contamination of a bacterial culture by mold (Alexander Fleming) Librium (1957, the first benzodiazepine): - Chemical synthesis mistake (accidentally used wrong reactant, made wrong compound) - Tested all but one anyway, just to see. No effect. - Decided to go back and test the last one a year and a half later just to see…

Answer 55

Secondary metabolites: Defenses against predators, parasites, diseases, interspecies competition, and to facilitate reproduction. - natural sources provide templates for drug innovation (different climate --> different relationships) - Natural products are often potents and have very specific biological activities (chemistry of molecules had eons to evolve for good affinity) - Isolating natural products requires extensive effort and expertise ** top hits are usually from plants ***34% of drugs approved from 1981-2010 were natural-product based

Answer 56

Difficult to ID Limited supply Difficult chemistry → manufacturing nightmare Expensive

Answer 57

Many drugs are entirely artificial, with no inspiration in natural compounds Chemical libraries → combinatorial chemistry Drug discovery begins by identifying and optimizing lead compounds

Answer 58

Bioassays: biological tests used to screen for bio activity/potency -- Key tools for screen compound libraries Activity: type of bio effect (antifungal, antibacterial…) Potency: magnitude of that effect

Answer 59

Bioassays: biological tests used to screen for bio activity/potency Key tools for screen compound libraries In vitro and in vivo bioassays offer complimentary screening insights

Answer 60

in vitro assay: Carried out in a test-tube Cheap and quick in vivo assay: Carried out in organisms (animals,plants) Expensive, but more reliable for testing activity and potency

Answer 61

Refines drug properties for efficacy and safety Synthesize and test hundreds (or thousands) of structurally related compounds Computational techniques can accelerate this process. “SAR by catalogue.” (structure activity relationship)

Answer 62

1. desalting 2. adding hydrogens 3. chirality 4. cis-trans isomerism 5. tautomers

Answer 63

Removing salts improves molecular database accuracy Salts: charged molecules on target Don't have bio activity, so remove -- Typically keep only the molecule with the greatest # of atoms

Answer 64

Yields models that better reflect biological reality -- SMILES strings don't include H-atoms Oxygen and nitrogen atoms can be more complicated because protonation depends on the pH of the solution and local protein environment ***use henderson-hasselbalch equation

Answer 65

Determines protonation states (predicts proton presence) More positive the pKa → weaker the acid Henderson-Hasselbalch equation: determines protonation given pH

Answer 66

determines protonation given a specific pH 10 ^ (pH - pKa) : 1

Answer 67

*** mirror-image, non-superimposable molecules Impacts molecular recognition SMILES can include information about chirality… But not always --- "@" means following neighbors are listed anticlockwise

Answer 68

Identical groups on the same VS opposite side of the double bond Affects biological activity

Answer 69

Interconvert based on environmental conditions structural isomers that rapidly interconvert through the movement of a proton Ex/ enol vs keto form

Answer 70

Force fields enable conversion from 2D to 3D structures approximates bonded and nonbonded forces (VDW and partial forces between atoms) Sets parameters for deciding ideal lengths/angles between atoms Bonds ⇒ springs (ideal length/tension with some flexibility)

Answer 71

1. initial atomic model 2. calculate molecular forces acting on an atom 3. move each atom according to those forces 4. advance sim time by 1-2 fs 5. repeat at step 2-4 to optimize

Answer 72

Capture the 3D shapes that molecules can adopt using: 1. Rotation about single bonds 2. Energy barrier 3. time in a given rotamer/interconversion between conformers Energy barriers define the stability/interconversion of conformers *** Some docking programs require a pre-calculated conformational library

Answer 73

Ring conformations depend on steric and electronic factors Some ring conformations are more energetically favorable/likely to have a higher retention time in this conformation Ex/ chair is lower in energy than boat

Answer 74

computer aided drug design uses QSAR models

Answer 75

Quantitative Structure-Activity Relationships Predicts activity without structural data Uses ML to predict if a molecule would bind to a target (w/o info on target structure) Looks for common features among drugs that all product the same biological response (design a drug that combines these features) Based on properties of small molecules only!!!

Answer 76

Same basic structure with varying substituents → provide insights for QSAR modeling Structure-activity relationships (SAR) Making minor changes in a lead compound to produce analogs

Answer 77

Dream: small changes always have small impacts on affinity Reality: Sometimes simple changes drastically alter binding (sometimes small changes prevent binding entirely)

Answer 78

Change the size/shape of carbon skeleton Change nature/degree of substitution Change the stereochemistry of the lead

Answer 79

Regression: continuous values of binding affinity Classification: does it bind? Yes or no?

Answer 80

Capture structural/chemical information (Physio-chemical properties) *** Molecular weight, LogP (lipophilicity), QM descriptors Careful selection of descriptors enhances QSAR accuracy Links properties to biological activity Goal is to map numerical properties of ligand to known measures of bio activity

Answer 81

0D: Molecular weight, number of bonds, tallies of atoms, etc. 1D: Derived from the molecular formula (atoms). Molecular fragment counts. Number of hydrogen bond donors and acceptors. 2D: Consider bonds between atoms/fragments. “Topological representation.” 3D: Considers shape of a small molecule in 3D space.

Answer 82

Advantages: Simple info about molecular structure. -- Don’t have to calculate the 3D conformation(s). Topostructural descriptors: Include information about the distance between atoms ”in bond space.” Topochemical descriptors: Topology as above, but also information about the chemical properties of the atoms --- chemical identity, hybridization

Answer 83

1. Chemical structures 2. Molecular properties (or you can calculate/predict from the structures) 3. Experimentally measured biological outcomes (e.g., binding affinities)

Answer 84

Training QSAR models minimizes prediction errors Linear regression is the “machine learning” method traditionally applied to QSAR. *** find the coefficients (ax) that minimize the error given descriptors xn and biological activity y.

Answer 85

1. The relationship between descriptors might not be linear. 2. You have to have a lot of known ligands. 3. You can only predict new ligands if they are similar to ones you trained on.

Answer 86

Ensures predictive accuracy -- Separate data into training and testing set N-fold cross-validation improves model reliability -- Changing what sections of the data are training/testing

Answer 87

Quantitative Structure-Property Relationships Predicts chemical properties *** BP, solubility, viscosity, carcinogenicity, drug metabolism and clearance

Answer 88

PoseView, which represents interactions in a 2D schematic BINANA provides a 3D perspective, helping you grasp the geometry of interactions between ligands and protein residues.