Fragment Libraries

The fragment-based approach to drug discovery (FBDD) has been established as an efficient tool in the search for new drugs [1,2]. It has emerged as an alternative approach to traditional lead identification via high-throughput screening (HTS). Unlike HTS, FBDD identifies smaller compounds, the “fragments”, which bind to different parts of a biological target. FBDD has several advantages over HTS. The primary rationale for screening fragments is that they give access to a broader chemical space while screening a limited number of compounds. FBDD gives a better chance for the final lead compound to have common drug-likeness parameters. Moreover, the chance of identifying a compound that is a perfect match for the intended target’s binding site is too small to rely on because a multitude of interactions between the ligand and amino acid residues in the active site are often involved. On the other hand, there are simpler molecules (fragments) with fewer unavoidable interactions and, thus, a much better chance of orienting within the binding site in a favorable manner.

In response to this evident trend in drug discovery Life Chemicals has designed its proprietary Fragment Library containing nearly 68,000 compounds.

Following the most promising current tendencies in fragment based approach to drug discovery, Life Chemicals has prepared a selection of its unique subsets given below:

In addition to these subsets the brochure presents our exclusive library of 5,200 tangible fragments – a product of unique design specifically developed for fragment-based drug discovery. The Library is a collection of first-rate quality fragments having a high fraction of sp3 carbons, notable diversity, exceptional novelty and rapid SAR generation potential. The structure files of our Fragment Libraries can be downloaded here.

GENERAL FRAGMENT LIBRARY

The Life Chemicals General Fragment Library comprises more than 68,000 compounds with MW ≤ 300 and ClogP ≤ 3.0, including 38,000 stock available compounds and 30,000 tangible structures.

More scrupulous assay applying expanded "Rule of Three" parameters and additional physicochemical and structural filters enabled us to work out our Advanced Set of nearly 9,200 stock-available fragments. The criteria used to design The Life Chemicals Advanced Fragment Set are listed in the table below. Additionally, the compounds were passed through Substructure[1] and PAINS[2] structural filters.

ParameterRangeAverage Values
MW 100 - 300 238
ClogP -2 - 3 1.0
TPSA < 100 Å2 62 Å2
Rotatable bonds ≤ 3 2.1
H-donors ≤ 3 1.2
H-acceptors ≤ 4 2.9
logS ≥ -3.5  
Rings count 1 - 4  
-CN, -NO₂, Br count ≤ 1  
S, Cl count ≤ 2  
Benzene count ≤ 1  


ULTIMATE FRAGMENT LIBRARY

The process of fragment synthesis obeys a heuristic rule called the “Rule of Three”. State-of-the-art tuning of “Rule of Three” parameters helps to generate leads with improved ADME profile.

The Life Chemicals Ultimate Fragment Library was designed by applying ultimately refined picking approach to our stock screening compound collection: the “Rule of Three” and TPSA ≤ 80 Å2 cut-off. It is known that more than 80 % of drugs on the market have an estimated logSw value greater than -4. Thus, the solubility filtering was also used to design this Fragment Subset.

Finally, PAINS and in-house developed filtering parameters, such as toxicophore and undesired functionalities, were applied to make up the set of compounds shown here. The total number of compounds in The Life Chemicals Ultimate Fragment Collection amounts to 4,200 items.

Physicochemical criteria used to create The Ultimate Collection are as follows:

ParameterRange
MW 150 - 300
ClogP -2 - 3
TPSA < 80 Å2
Rotatable bonds ≤ 3.0
H-donors ≤ 3.0
H-acceptors ≤ 3.0
ClogSw -3
Halogens (except F) ≤ 1
S atoms ≤ 1
S atoms ≤ 1
Ring count 1 - 3
Fused rings ≤ 2
Benzene count ≤ 1


LOW MW FRAGMENT LIBRARY

Recent tendencies in drug discovery, however, shift towards aiming at leads with lower molecular weight and higher hydrophilicity; it is not surprising, therefore, that even more meticulous criteria must be applicable to fragment libraries. In particular, lowering upper molecular weight cut-off value (to 200–240 instead of 300) is strived for by many companies practicing FBDD.

To meet these principles, Life Chemicals has designed its Low MW Fragment Library applying a number of physicochemical filters [1, 2] to its General Fragment Collection. Finally, the compounds were passed through MedChem [3] and PAINS [4] structural filters. This resulted in a library of more than 3,000 compounds. Although the molecular weight is the main parameter which was controlled strictly, other physico-chemical characteristics were also well monitored (e. g., more than 80 % of the library has ClogP<2). Due to this upgrading a refined set of nearly 1,500 fragments was generated by means of more rigorous physicochemical and structural cut-offs.

ParameterRangeAverage Values
Low MW Fragment LibraryAdvanced Low MW Fragment Set
MW 100 - 225 177 186
ClogP -3 - 3 0.7 0.9
TPSA 0 - 100 Å2 55 Å2 47 Å2
Rotatable bonds ≤ 2 1.3 1.3
H-donors ≤ 3 1.3 1.3
H-acceptors ≤ 6 2.5 2.1
Ring count 1 - 3 1.8 2.0


FRAGMENT LIBRARY WITH EXPERIMENTAL SOLUBILITY DATA

Solubility of fragments is known to be one of the key features to limit their use in various screening techniques of FBDD. Proceeding from this knowledge, Life Chemicals has started its project on measuring experimental solubility of proprietary fragment collection. The procedure includes visual determination of solubility by observing scattering of solutions under the following concentrations:

  • 200 mM in DMSO
  • 5 mМ in water phosphate buffer with 2.5 % DMSO
  • 1 mМ in water phosphate buffer with 0.5 % DMSO

The Life Chemicals Fragment Library with Experimental Solubility Data includes over 12,800 compounds, mainly from our Advanced Set of The General Fragment Library, with confirmed experimental solubility at 200 mM in DMSO and characterized by the data available for the above mentioned concentrations in phosphate buffer.

  • Average MW 240 Da and logP 1.2
  • Guaranteed solubility of all compounds in DMSO at high concentration (200 mM)
  • 77 % of the Library are soluble in water phosphate buffer at 1 mM, and 60 % – at 5 mM
  • Compounds were passed through MedChem structural filters worked out by our experts


Fsp3-ENRICHED FRAGMENT LIBRARY

According to new trends in drug discovery involving comprehensive structural analysis of approved drugs and determination of relationships between molecular complexity and pharmacological promiscuity of drug-like compounds [1–5], Life Chemicals has extrapolated the results of investigations described here to its Fragment Library design.

Recently it was found that the mean Fsp³ goes up from 0.36 for 2.2 million molecules in drug discovery phase to 0.47 for 1,179 approved drugs. Applying the mentioned above Fsp³ cut-off to The Life Chemicals General Fragment Library resulted in our Fsp³-enriched Fragment Library of more than 11,500 compounds.

Superposition of sets of the compounds with a relatively high Fsp³ value and filtering them by strict “Rule of Three” criteria along with TPSA 90Å2 cut-off gives rise to our Advanced Fsp³-enriched Fragment Subset comprising nearly 3,200 compounds. To ensure discarding “ugly” compounds, PAINS and our in-house developed toxicophore and undesired functionalities filtering were applied as well.

Physicochemical parameters of The Advanced Fsp³-enriched Fragment Library are summarized in the table below:

ParameterRangeAverage Values
MW ≤ 300 224
ClogP < 3 0.96
Fsp³ > 0.45 0.65
TPSA < 90 Å2 52 Å2
Rotatable bonds ≤ 3 2.2
H-donors ≤ 3 1.3
H-acceptors ≤ 3 2.2
Benzene count ≤ 1 0.5


BROMINE FRAGMENT LIBRARY

Crystallographic fragment-based drug discovery is gaining a high importance in modern pharmaceutical technology. Fragments, however, are poor binders, with their partial occupancy resulting in weak, difficult-to-fit electron density. The use of brominated fragments can address this challenge, as bromine can be located uniquely via anomalous scattering, and thus, provide further insight into the nature of the target surface pocket [1].

The Life Chemicals Bromine Fragment Library consists of about 1,300 compounds with MW ≤ 350 and ClogP ≤ 3.5. All the fragments contain only one Br-atom. Other important parameters (such as HBA, HBD, Fsp3, RotB, TPSA) are included into the data file and can be used for selection with multiple restrictions. While this library is available for regular cherry picking, a specially tailored diversity set can be arranged in accordance with the customer’s needs.

A subset of more than 800 compounds composed on the basis of the values given below is also available where there are no reactive compounds and undesired chemotypes, with PAINS filters being applied.

ParameterRange
MW < 350
ClogP < 3.5
Fsp³ > 0.4
Number of rings 1 - 3
Rotatable bonds ≤ 4
H-donors 0 - 3
H-acceptors 0 - 4


FLUORINE FRAGMENT LIBRARY

It was demonstrated that ¹⁹F NMR-based fragment screening can be used as a very efficient tool for rapid and sensitive detection of fragment hits [1]. As shown, although the fraction of chemical space covered by fluorine containing fragments may be limited compared to that covered by non-fluorinated fragments, similar hit rates are obtained applying both methods. At the same time fragment screening with a simple onedimensional ¹⁹F NMR experiment (with ¹H decoupling) is significantly faster and, in many ways, more robust than traditional ¹H-based NMR screening. An in-house developed method focuses on the use of a fluorine nucleus as a detection tool, and subsequent to a ¹⁹F fragment screen, protonated analogues of hit compounds can be mined and pursued in initial hit expansion. This approach can significantly add to the chemical space explored and provide valuable information and early stage SAR to chemists.

The Life Chemicals Fluorine Fragment Library extracted from The General Fragments Stock Collection comprises about 3,300 fluorine-containing compounds, including an Advanced Subset of nearly 2,000 compounds selected by the following criteria:

ParameterRange
MW < 350
ClogP < 3.5
Fsp³ > 0.4
Number of rings 1 - 3
Rotatable bonds ≤ 4
H-donors 0 - 3
H-acceptors 0 - 4
Number of acidic groups < 2
Number of basic groups < 2

No metal or reactive compounds, no undesired chemotypes, no salts, PAINS filters applied

“Fragment space” coverage of Life Chemicals fluorinated fragments (in black) is similar to the coverage of all commercially available fluorinated fragments (in green).

FLUORINE FRAGMENT COCKTAILS

The efficacy of ¹⁹F NMR fragment screening can be increased substantially by pooling or cocktailing the compounds in the library. Identification of the bound fragment at the end of the NMR experiments, then, becomes the case of determining the best fragment-fit.

Life Chemicals has designed a unique Library of Fluorine Fragment Cocktails, employing “Rule of Three” criteria and MedChem structure filters, including PAINS. The choice of fluorinated motifs present in the library is fundamental in order to ensure a larger coverage of chemical space and local environment of fluorine. The Library of 1,010 fluorinated fragments is pooled in 101 cocktails of 10 compounds, each with most different ¹⁹F chemical shifts to facilitate interpretation of screening results.

The fragments chosen for cocktails preparation satisfy the following parameters:

  • Mostly only 1 peak in ¹⁹F NMR spectrum
  • Purity 95 %+
  • DMSO soluble at 200 mM

Structure Filters Applied:

  • No reactive functional groups
  • Lilly MedChem rules
  • PAINS
  • Ring count 1 – 3

Physicochemical Parameters:

  • 130 < MW < 300
  • -1.7 < ClogP < 3.0
  • H-bond Donors ≤ 3
  • H-bond Acceptors ≤ 4
  • TPSA ≤ 110 Å2

¹⁹F NMR spectrum of fragment cocktail # 101 from The Life Chemicals Library of Fluorine Fragment Cocktails.



PPI FRAGMENT LIBRARY

Protein-protein interactions (PPIs) are among the most attractive targets for drug design. However, this field of interactions is not well explored and there are no common features for all proteins involved in it. Several dozens of protein pairs have been fully investigated so far and their crystal structures have revealed the mechanism of tight connection between interfaces. These investigations present a hard task consuming too much time and effort but they can be facilitated by development of fragment-based library focused on inhibition of protein-protein interactions.

A set of descriptors and ranking mechanism of arraying fragments was formulated on the basis of our thorough analysis of published data that enabled us to construct a new Life Chemicals PPI Targeted Fragment Library.

ParameterRangeAverage Values
MW 150 - 450 345
ClogP 2.5 - 4.5 3.3
Fsp³ > 0.4 0.51
TPSA < 75 Å2 58.9 Å2
Rotatable bonds ≤ 8 4.6
H-donors ≤ 3 1.1
H-acceptors ≤ 5 2.9
Hyrophobic cores 1 - 6 2.6

It was shown that PPI inhibitors are typically larger and more lipophilic than inhibitors of more standard binding sites of most proteins [1]. That’s why PPI fragments have TPSA values higher than it was generally claimed when practicing a common approach to fragment-based drug design. The same applies to molecular weight of compounds that should be between 250 and 450 Da. Thus, such compounds cannot fit a standard “Rule of Three” usually used for selection of fragments. It was found that small molecules targeting PPI have highly hydrophobic core within peripheral lipophilic substituents [2]. To enhance selectivity we have proposed very reasonably chosen hydrophobic and more spatial structures (sp³-enriched).

The main result of compilation of the features mentioned above is a special chemical space for compounds with a common function of PPI interaction inhibition. One more detail that characterizes PPI fragments is the principal moment of inertia (PMI). Mean values of npr1 = 0.28, npr2 = 0.88 show the distribution of compounds in 3D space and evaluate 3D-shape diversity [3]. Finally, the compounds were passed through PAINS filter.

More than 2,500 compounds were extracted into The Life Chemicals PPI Fragment Library.

3D FRAGMENT LIBRARY

The vast majority of existing drugs has centered on sp2-rich aromatic core [1,2]. Moreover, we could often find the same core fragment in several drugs with different targeted diseases that leads to a low specificity (selectivity) and the rise of side effects. Although most fragment libraries are diverse, having been refined to contain a good balance of properties, they all tend to have a limited shape diversity [5]. In this context it has been established that a higher three-dimensionality (3D) of molecules is a desirable feature of drug candidates and is correlated with successful passage of molecules at various stages of clinical development. The use of more complex and sp3-rich (more 3D-like) fragments would undoubtedly build up the drug-like fragment chemical space that might in turn be advantageous in exploring more demanding biological targets.

In view of the above, we have carefully designed our drug-like 3D Fragment Library using physicochemical properties and descriptors that allow to evaluate 3D-dimensionality and diversity of the molecules. The “Rule of Three” with a number of filtering criteria enabling successful 3D library design was applied to The Life Chemicals Stock Fragment Collection. Principal moments of inertia (PMI) [3,6] calculation was used as an efficient method to calculate and evaluate 3D-dimensionality. Proper diversity levels of the library were proved by applying max Tanimoto coefficient of diversity 85 % (linear fingerprints were used). Undesirable functionalities were eliminated by applying PAINS and our special in-house medicinal chemistry filters. Totally 1,500 fragments were chosen. The following basic criteria were used to improve 3D shape functionality of fragments in our library:

ParameterRange
MW 100 - 300
Fsp3 > 0.47
TPSA < 100 Å2
Rotatable bonds ≤ 3.0
H-donors ≤ 3.0
H-acceptors ≤ 4.0
Chiral centres ≥ 1
Functionalization points 2
-CN, -NO2, Br count ≤ 1
S, Cl count ≤ 2
Ring count 1 - 4
COVALENT FRAGMENT LIBRARY

In recent years, the number of drug candidates with a covalent mechanism of action progressing through clinical trials or being approved by the FDA has grown significantly; around 30 % of the marketed drugs are covalent inhibitors. This kind of inhibition has many desirable features, including higher biochemical efficiency of target disruption, less sensitivity toward pharmacokinetic parameters and increased duration of action that outlasts the pharmacokinetics of the compound [1].

Taking into account a growing interest and widespread use of fragment-fased drug discovery, Life Chemicals designed the Library of 1,300 Covalent Fragments based on The Stock Compound Collection. The design of selective covalent inhibitors is conceptually very attractive but in practice hard to achieve. That is because it is difficult to strike the right balance between reactivity and selectivity [2-4]. Initially, The Life Chemicals Stock Collection was filtered according to “Rule of Three” criteria. Thereafter, a preliminary set of covalent fragments was created by singling out compounds with specific structural fragments (functional groups) [4-8] that are known to form covalent bonds with amino acid residues in binding sites of targeted proteins, e.g. Lys, Cys, Ser, His and Tyr. In addition, moleculeswith highly reactive electrophilic and nucleophilic groups (non-selective binders) as well as compounds with non-drug-like cores were discarded, using unwanted fragments and appropriate PAINS filters.

Chemical classes and structural fragments shown below were used for arraying potential covalent inhibitors:

  • Acetals;
  • Acrylamides;
  • Acrylonitriles;
  • a-Dicarbonyl compounds;
  • Aliphatic thiols;
  • Aromatic thiols;
  • Epoxides;
  • Ketals;
  • Maleimides and related compounds;
  • Other Michael acceptors;
  • Quinones;
  • Sulfonate esters;
  • Sulfonyl halides;
  • Terminal acetylenes;
  • Thioureas and thiones;
  • Vinyl sulfones;
  • Vinyl sulfonamides;
  • Other types.


NATURAL PRODUCT-LIKE FRAGMENT LIBRARY

A remarkable structural diversity and drug-likeness of molecular scaffolds identified in natural compounds provides a basis for the design of novel natural product-derived compound libraries for drug discovery [1,2]. Implementation of natural product-like frameworks into fragment-based drug discovery might benefit from using biologically validated areas of chemical space not occupied by average synthetic molecules [3,4].

Life Chemicals presents a collection of Natural Product-like Fragments generated via Scaffold Tree approach [5]. More than 17,000 level-two and level-three scaffolds were extracted from The Universal Natural Product Database (comprising more than 220,000 structures in its current version) [6] using scaffold tree analysis. After excluding undesirable and primitive chemotypes, the remaining ca. 4,000 structures were used for substructure search within The Life Chemicals Fragment Collection. In addition to that, fragments derived from scaffolds showing at least 85 % similarity with the natural product-derived scaffold set mentioned above were included as well. The resulting database was subject to additional structural filtering to give more than 3,200 natural product-like fragments.