Skip to main content
NCBI home page
ACS Omega logoLink to ACS Omega
. 2025 Apr 16;10(16):16921–16937. doi: 10.1021/acsomega.5c01420

Fragment Libraries from Large and Novel Synthetic Compounds and Natural Products: A Comparative Chemoinformatic Analysis

Verónica Ramírez-Cid †, Ana L Chávez-Hernández †, Osvaldo Sánchez López †, Raul Marques Novais ‡, Temitayo Omowumi Alegbejo Price ‡, Kamilla Moraes Alves ‡,§, Wemenes J Lima Silva ‡,§, Flavio da Silva Emery ‡, Carolina Horta Andrade ‡,§,∥,*, José L Medina-Franco †,*
PMCID: PMC12044453  PMID: 40321506

Abstract

graphic file with name ao5c01420_0011.jpg

We report comprehensive fragment libraries obtained from large natural product databases and compare their chemical space coverage and diversity with those of synthetic fragment libraries. Specifically, we obtained 2,583,127 fragments derived from the recently updated collection of open natural product (COCONUT) data set with more than 695,133 unique (nonduplicate) natural products and 74,193 fragments derived from the Latin America Natural Product Database (LANaPDB) with 13,578 unique natural products from Latin America. The content, chemical space coverage, and chemical diversity of the natural product libraries were compared to the recently developed CRAFT library, which contains 1214 fragments based on distinct heterocyclic scaffolds and natural product-derived chemicals. The fragment libraries herein obtained and curated are freely available at https://github.com/DIFACQUIM/Fragment-libraries-from-large-synthetic-compounds-and-natural-products-collections.git.

1. Introduction

Major sources of small molecule drugs are natural products (NPs) and chemical synthesis (mostly organic synthesis). Historically, humans have used NP, particularly plants, in the treatment of different diseases.1−3 An overall challenge developing NP-based drugs is their structural complexity that is frequently associated with the difficulty to synthesize them. Continued efforts have been made to design compounds that resemble NP and are rich in sp3 atoms.4−6 An approach is to deconstruct NP into fragments and combine unrelated NP fragments into pseudo-NP using synthetic methods.5,7,8

Over the past few years, efforts have been made to compile a list of the NPs. For instance, the second version of the COlleCtion of Open Natural prodUcTs (COCONUT) has 695,133 distinct structures.9,10 Similarity, Latin American countries have been assembling a unified and open-access Latin American Natural Product Database (LANaPDB) that gathers ten databases.11 The first version of LANaPDB had 12,959 chemical structures12 and the most recent version has 13,578 compounds.13

The integration of fragment-based drug design (FBDD) with other drug discovery approaches has become crucial in the development of new candidates. The increasing application of biophysical and biochemical techniques for fragment identification, alongside rational library design, facilitates the transformation of fragments into promising chemical series. The significance of FBDD underscores its vital role in fostering innovative therapies and broadening the horizons of medicinal chemistry.14,15

FBDD typically utilizes small organic molecules with fewer than 20 non-hydrogen atoms, adhering to the “rule of three” (RO3), ensuring a more efficient exploration of chemical space.14,16 Sources such as NPs and synthetic libraries17 further enhance this process by providing fragments with ideal physicochemical properties for the creation of lead series.

Unlike HTS, which screens hundreds of thousands of molecules, FBDD relies on smaller libraries, typically comprising 1000 to 5000 low-molecular-mass fragments. This enables more efficient screening, reduced operational costs, and greater practicality in library management. Over the past two decades, FBDD has achieved remarkable milestones, including the approval of drugs (for example: venetoclax,18 vemurafenib,19 sotorasib,17 and pexidartinib20) and the development of over 40 compounds in clinical stages originating from FBDD screenings. This success underscores its significance in addressing various diseases.21,22

For computer-guided FBDD, fragments can be obtained through deconstruction methods such as REtrosynthetic Combinatorial Analysis Procedure (RECAP),23 Breaking of Retrosynthetically Interesting Chemical Substructures (BRICSs),24 and MOlecule fRagmenTAtion fRamework (MORTAR),25 a framework integrated with three algorithms: ErtlFunctional GroupsFinder,26 Sugar Removal Utility, and Scaffold Generator.27 Fragmentation algorithms identify characteristic molecular structures in a comprehensible way based on structural definitions like functional groups; therefore, the main advantage of fragmentation algorithm RECAP regarding MORTAR algorithms (scaffold generations or group finder) is that a fragment captures information both molecular scaffolds and functional groups.25

Another method of obtaining molecular fragments for computer- and experimentally guided FBDD is from commercial vendors or academic groups. Commercial vendors have collections of this type of building blocks, for example, Enamine (12,000 fragments),28 ChemDiv (74,000 fragments),29 Maybridge (30,000 fragments),30 and Life Chemicals (65,000 fragments).31

The center for research and advancements in fragments and molecular targets (CRAFT) has made available its synthetic database of compounds and fragments at https://github.com/CRAFT-Therapeutics/Fragment-library.git. CRAFT is a Brazilian interinstitutional collaboration between the University of Sao Paulo and the Federal University of Goias. A major goal of CRAFT is to advance drug discovery, with a special emphasis on neglected infectious diseases as well as emerging diseases such as bacterial and viral infections.32 CRAFT has compiled a fragment library containing structures based on new heterocyclic scaffolds and compounds derived from NPs.33 All the fragments in the CRAFT library were obtained experimentally, and this plays a main role in lead discovery34 because some computer-designed compounds cannot be synthesized.35 Molecules generated from the CRAFT’s fragments are designed to be synthetically accessible.32

The goal of this study was to analyze the contents, properties, and chemical diversity of fragment libraries obtained from two large natural product libraries (COCONUT 2.0 and the most updated version of LANaPDB), chemical synthesis (CRAFT), and commercial fragment libraries (Enamine, ChemDiv, Maybridge, and Life Chemicals).

2. Methodology

2.1. Data Sets

Table 1 summarizes the data sets used in this work. COCONUT is a large library with the chemical structures and annotations of 695,133 NPs and is a compendium of other publicly available NP collections including LANaPDB. The latest update of LANaPDB has 13,578 nonduplicate NPs. Table 2 summarizes the number of compounds in CRAFT and the commercial and synthetic fragment libraries. CRAFT’s fragments contain 1214 fragments obtained by chemical synthesis. Commercial fragment libraries were 12,505 Enamine’s fragments with solubility in water, 74,721 ChemDiv’s fragments, 30,099 Maybridge’s fragments, and 65,552 Life Chemical’s fragments.

Table 1. Natural Product Libraries Studied in This Work.

data set initial size size after standardization protocol initial number of fragmentsa fragments after standardization protocol fragments that fulfill the RO3 (percentage) refs
LANaPDB 13,578 13,578 74,193 74,193 1832 (2.5) (13,36)
COCONUT 695,133 648,721 2,583,127 2,583,127 38,747 (1.5) (9,10)
Open in a new tab
a

Compounds with molecular weight larger than 1000 Da were excluded.

Table 2. CRAFT Fragment Library and Commercial Fragment Libraries from Chemical Vendors.

data set initial fragments number of fragments after standardization protocol fragments that fulfill all properties of the RO3 (percentage) refs
CRAFT 1214 1202 176 (14.6) (32)
Enamine (soluble in water) 12,505 12,496 8386 (67.1) (28)
ChemDiv 74,721 72,356 16,723 (23.1) (29)
Maybridge 30,099 29,852 5912 (19.8) (30)
Life Chemicals 65,552 65,248 14,734 (22.6) (31)
Open in a new tab

2.2. Data Set Standardization

Fragment libraries were stored using simplified Molecular Input Line Entry System (SMILES) strings.37 The fragments were prepared and curated using toolkits RDKIT version (2024.03.5)38 and MolVS version 0.1.1.,39 and standardization protocol was described by Sánchez-Cruz et al.40 Fragments were selected if they had the elements H, B, C, N, O, F, Si, P, S, Cl, Se, Br, and I. Fragments with multiple components were split, retaining the largest component. Then, fragments were reionized and neutralized, and a canonical tautomer was generated. Finally, unique fragments were retained.

2.3. Fragment Libraries

NP fragment libraries were generated for the compounds in COCONUT10 and LANaPDB13 with molecular weight less than 1000 Da. This threshold was selected to guarantee the fragmentation of at least 95% of the compounds, also because fragmentation of larger molecules takes significantly longer. The fragments were obtained using the RECAP23 function from the RDKit toolkit. RECAP breaks eleven chemical bonds as amine, amide, ester, urea, olefin, ether, aromatic nitrogen–aliphatic carbon, lactam nitrogen–aliphatic carbon, aromatic carbon–aromatic carbon, quaternary nitrogen, and sulfonamide. The synthetic fragment library was obtained from CRAFT (available at https://github.com/CRAFT-Therapeutics/Fragment-library.git).32 We also gathered four commercial fragment libraries, summarized in Table 1.

Fragments fulfilling the RO316 were retained and are referred to in this manuscript as “Fragment RO3”. The RO3 rule describes fragments that have six properties of pharmaceutical relevance: molecular weight (MW), rotatable bonds (RBs), topological polar surface area (TPSA), partition coefficient octanol/water (Log P), hydrogen-bond acceptors (HBAs), and hydrogen-bond donors (HBDs), with the values: MW ≤ 300 Da, RB ≤ 3, TPSA ≤60 Å2, logP ≤3, HBA ≤3, and HBD ≤3.

2.4. Synthetic Accessibility Score

The synthetic accessibility score (SA score) is a calculated value that approximates the feasibility of a molecule to be synthesized.34 SA score is calculated by the difference between fragment score and complexity penalty. The fragment score indicates the structural feature viability of synthesized molecules and is calculated as a sum of contributions of all fragments contained in the molecule. The complexity score is the sum of ring complexity (ring bridge atoms and spiro atoms), large rings, stereocenters, and molecular size.41,42 The SA score was computed using the Python script of Ertl and Schuffenhauer.34

2.5. Content, Complexity, and Structural Diversity

Fragments and “Fragment RO3” were analyzed using fourteen constitutional and complexity descriptors (cf. Table 6). Their structural diversity was measured using the Tanimoto coefficient43 and the following fingerprints: Molecular ACCes System (MACCS) keys (166 bit)44 and Morgan fingerprints45 with radius 2 (Morgan2, 1024 bit) and radius 3 (Morgan3, 1024 bit).

Table 6. Constitutional and Complexity Descriptors of NPsa.

data set COCONUT LANaPDB
carbon atoms 28.27 21.34
oxygen atoms 6.6 5.43
nitrogen atoms 1.26 0.21
fraction of carbons 0.78 0.8
fraction of oxygens 0.18 0.19
fraction of nitrogens 0.04 0.01
fraction of sp3 carbons 0.57 0.59
fraction of chiral carbons 0.18 0.21
molecular weight 510.91 374.69
heavy atoms 36.35 27
rings 3.74 3.62
aliphatic rings 2.35 2.64
aromatic rings 1.39 0.98
heterocycles 1.62 1.27
aliphatic heterocycles 1.19 1
aromatic heterocycles 1.39 0.98
spiro atoms 0.18 0.35
bridgehead atoms 0.47 0.55
Open in a new tab
a

Mean value of the distribution.

2.6. Chemical Space and Chemical Multiverse Analysis

Chemical space can be defined as an M-dimensional Cartesian space, each of the dimensions encoding a molecular descriptor of a set of molecules.46 The number of descriptors describes the number of dimensions that make up this chemical space, and the type or nature of the descriptors defines the specific type of chemical space (e.g., property-based and fingerprint-based chemical space). The chemical multiverse concept is a natural extension of the chemical space and it can be defined as a group of alternative chemical spaces of a set of molecules, each defined by a different set of molecular descriptors.47 In this study, the chemical space visualization was done using the fingerprints Morgan2 (1024 bit), Morgan3 (1024 bit), and MACCS keys (166 bit) and two algorithms: Tree MAP (TMAP)48 and T-distributed Stochastic Neighbor (t-SNE).49 TMAP is grouped hierarchically as compounds according to its common structures using a molecular fingerprint, nearest-neighbor algorithm (k), and query algorithm (kc). TMAP was generated using k = 50 and kc = 10. t-SNE was generated using the number of nearest neighbors (perplexity = 40) and the number of iterations (n_iter = 300).

3. Results and Discussion

Table 1 summarizes the number of compounds before and after the standardization protocol (Section 2.2) and the number of fragments generated for the two NP libraries. It is noteworthy that there is a large number of fragments generated from COCONUT (more than 2.5 million). For both NP libraries, the percentage of “Fragment RO3” is relatively small (2.5% and 1.5% for LANPDB and COCONUT, respectively). Table 2 summarizes the number of total fragments and “Fragment RO3” for CRAFT and the four fragment libraries of chemical vendors. Enamine, ChemDiv, and Life Chemicals had a higher percentage of “Fragment RO3” (67.1%, 23.1%, and 22.6%, respectively) than Maybridge and CRAFT (19.8% and 14.6%, respectively). Not surprisingly, the proportion of fragments that fulfill the RO3 in the chemical vendors (between 22.6% and 67.1%) is higher than the same type of fragments in NP libraries (1.5% and 2.5%).

3.1. Unique Natural Product Fragments

Figures 1 and 2 show the ten most frequent chemical structures from LANaPDB’s fragments, LANaPDB’s “Fragment RO3” (Figure 1), and COCONUT’s fragments and COCONUT’s “Fragment RO3” (Figure 2). The percentage of each fragment is indicated below for each chemical structure. LANaPDB’s fragments (Figure 1A) had the most fused bicycles (twelve different and unique fragments with around 0.02% and 0.01%) and bridged bicycles compared with COCONUT’s fragments.

Figure 1.

Figure 1

Open in a new tab

Chemical structures were done with Marvin 17.21.050 of the twenty most frequent and unique (A) LANaPDB’s fragments and (B) LANaPDB’s “Fragment RO3”. The percentage of each fragment is indicated below each structure.

Figure 2.

Figure 2

Open in a new tab

Chemical structures of the twenty most frequent and unique (A) COCONUT’s fragments and (B) COCONUT’s “Fragment RO3”. The percentage of each fragment is indicated below each structure.

Fragments fulfilling RO3 showed noticeable differences. For example, LANaPDB’s “Fragment RO3” (Figure 1B) had the most fused bicycles and macrocycles (sixteen different fragments with 0.01% each one and around ten atoms). LANaPDB’s fragments had more bridged bicycles (three different and unique fragments with around 0.01%) than LANaPDB’s “Fragment RO3”.

The most frequent fragments from COCONUT (Figure 2A) are aliphatic compounds, mostly esters (nine fragments with 0.04%–0.05%) and ketones (five fragments with 0.03%). In contrast, the chemical structures of the twenty most frequent “Fragment RO3” obtained from COCONUT (Figure 2B) had a larger diversity and contained aliphatic alcohols (3.09–2.95%), aromatic rings (0.21%), bridge bicycles (0.14%–0.15%), and aliphatic cycles (0.17%–0.11%). It is noteworthy the high number of oxygen-containing fragments and the presence of very small fragments or mini fragments.

3.2. Common Fragments between Natural Product Libraries

Figure 3 and 4 show the forty most frequent fragments and “Fragment RO3” between COCONUT and LANaPDB. Figures 3 and 4 consider fragments with MW between 70 and 300 Da. The percentage of each fragment is indicated below each chemical structure. The first value (top) is for COCONUT and the second (bottom) is for LANaPDB. Figure 3 shows that the common structures between COCONUT and LANaPDB’s fragments were tetrahydropirans (0.23% and 0.28%), toluene (0.14% and 0.17%), benzene (0.13% and 0.18%), phenol (0.1% and 0.2%), anisol (0.08% and 0.11%), benzopyrans (0.04% and 0.14%), and nicotinaldehyde (0.02% and 0.05%). Figure 4 indicates that the common “Fragments RO3” between COCONUT and LANaPDB (Figure 4) were tetrahydropirans (0.24% and 0.42%), ophiocerin B (0.07% and 0.02%), nicotinaldehyde (0.11% and 0.42%), piperidine (0.11% and 0.2%), 1,3-benzodioxole (0.1% and 0.15%), and derivatives of 3-methoxyphenyl (0.07% and 0.23%).

Figure 3.

Figure 3

Open in a new tab

Forty most frequent and common fragments between the fragment libraries of LANaPDB and COCONUT generated in this study. Fragments with MW between 70 and 300 Da were considered. The percentage of each fragment is indicated below each chemical structure (proportion in COCONUT is the number on top). Marvin 17.21.0 was used for drawing the chemical structures.50

Figure 4.

Figure 4

Open in a new tab

Forty most frequent and common fragments between the “Fragment RO3” libraries of LANaPDB and COCONUT generated in this study. Fragments with MW between 70 and 300 Da were considered. The percentage of each fragment is indicated below each chemical structure (proportion in COCONUT is the number on top). Marvin 17.21.0 was used for drawing the chemical structures.50

Tables 3–5 summarize the number of unique and common fragments and “Fragment RO3” for the CRAFT, LANaPDB, and COCONUT libraries, as compared to the reference fragment collections.

Table 3. Unique and Overlapping Fragments between CRAFT and the Reference Libraries.

data set COCONUT LANaPDB Enamine ChemDiv Maybridge Life Chemicals
unique fragments present in the CRAFT’s fragment librarya 1139 (94.76%) 1199 (99.75%) 1195 (99.42%) 1152 (95.84%) 1179 (98.09) 1178 (98.00%)
overlapping structures with CRAFT’s fragments 63 (5.24%) 3 (0.25%) 7 (0.58%) 50 (4.16%) 23 (1.91%) 24 (2.00%)
unique fragments present in the CRAFT’s “Fragment RO3”b 144 (81.82%) 175 (99.43%) 171 (97.16%) 153 (86.93%) 163 (92.61%) 168 (95.45%)
overlapping structures with CRAFT’s “Fragment RO3” 32 (18.18%) 1 (0.57%) 5 (2.84%) 23 (13.07%) 13 (7.39%) 8 (4.55%)
Open in a new tab
a

1202 CRAFT’s fragments.

b

176 CRAFT’s “Fragment RO3”.

Table 5. Unique and Overlapping Fragments between COCONUT and the Reference Libraries.

data set CRAFT LANaPDB Enamine ChemDiv Maybridge Life Chemicals
unique fragments present in the COCONUT’s fragmentsa 2,583,064 (99.997%) 2,547,749 (98.63%) 2,582,782 (99.99%) 2,580,338 (99.89%) 2,581,975 (99.96%) 2,582,240 (99.97%)
overlapping fragments with COCONUT 63 (0.003%) 35,378 (1.37%) 345 (0.01%) 2789 (0.11%) 1152 (0.04%) 887 (0.03%)
unique fragments present in the COCONUT “Fragment RO3”b 38,715 (99.92%) 37,374 (96.46%) 38,497 (99.35%) 37,171 (95.93%) 38,014 (98.11) 38,231 (98.67%)
overlapping fragments with COCONUT RO3 32 (0.08%) 1373 (3.54%) 250 (0.65) 1576 (4.07%) 733 (1.89%) 516 (1.33%)
Open in a new tab
a

2,583,127 COCONUT’s fragments.

b

38,747 COCONUT’s “Fragment RO3”.

Table 3 indicates that CRAFT has a high percentage of unique fragments (94.76%–99.75%) and “Fragment RO3” (81.82%–99.43%) as compared to COCONUT, LANaPDB, ChemDiv, Enamine, Maybridge, and Life Chemicals. COCONUT (63, 5.24%) was the database with the highest percentage of fragments in common with CRAFT. Similarly, CRAFT and COCONUT have 32 (18.18%) “Fragment RO3” in common. In contrast, LANaPDB shared the fewest structures with CRAFT. CRAFT has a high percentage of 99.75% (1199) fragments and 99.43% (175) “Fragment RO3” not present in LANaPDB. Only 1 “Fragment RO3” is common between LANaPDB and CRAFT.

Table 4 shows that LANaPDB has a high number of fragments (47.68%) and “Fragment RO3” (74.95%) in common with COCONUT, which is to be expected since COCONUT integrates the first version of LANaPDB. With respect to the other reference libraries, LANaPDB maintains more than 99.9% of its fragments and more than 98% of “Fragment RO3” unique.

Table 4. Unique and Overlapping Fragments between LANaPDB and the Reference Libraries.

data set COCONUT CRAFT Enamine ChemDiv Maybridge Life Chemicals
unique fragments present in the LANaPDB’s fragmentsa 38,815 (52.32%) 74,190 (99.996%) 74,167 (99.96%) 74,147 (99.94%) 74,149 (99.94%) 74,172 (99.97%)
overlapping fragments with LANaPDB’s fragments 35,378 (47.68%) 3 (0.004%) 26 (0.04%) 46 (0.06%) 44 (0.06%) 21 (0.03%)
unique fragments present in the LANaPDB’s “Fragment RO3”b 459 (25.05%) 1831 (99.95%) 1809 (98.74%) 1803 (98.42%) 1802 (98.36%) 1815 (99.07%)
overlapping fragments with LANaPDB’s “Fragment RO3” 1373 (74.95%) 1 (0.05%) 23 (1.26%) 29 (1.58%) 30 (1.64%) 17 (0.93%)
Open in a new tab
a

74,193 LANaPDB’s fragments.

b

1832 LANaPDB’s “Fragment RO3”.

Table 5 reports the number of unique and overlapping fragments between COCONUT and the reference libraries. The percentage of unique fragments is in the range of 98.63%–99.99% and “Fragment RO3” is between 95.93% and 99.92%. COCONUT had the highest number of fragments in common with LANaPDB (35,378). However, ChemDiv has 1576 (4.07%) “Fragment RO3” in common with COCONUT, even more than those between LANaPDB and COCONUT (1373 “Fragment RO3”, 3.54%).

On the other hand, the fraction of NP “Fragment RO3”, LANaPDB RO3, and COCONUT RO3, present in fragment libraries commercially available from vendors (Enamine, ChemDiv, Maybridge, and Life Chemicals), was low. For instance, LANaPDB “Fragment RO3” (0.93%–1.64%) and COCONUT “Fragment RO3” (0.65%–4.07%) were low regarding to CRAFT’s “Fragment RO3” (2.84%–13.07%). The fragments obtained experimentally by both institutes or commercial vendors can support the lead discovery.34

The number of unique fragments indicates a higher degree of structural diversity. The NP databases are diverse, with respect to the reference databases. Similarly, CRAFT shows diversity, although it shares more fragments and “Fragment RO3” in common with COCONUT than with other libraries. The reason for this could be that overlapping the fragments that comprise CRAFT are NP derived.32 This diversity is clearly reflected in CRAFTs’ library, which incorporates a wide array of structural frameworks derived not only from natural and semisynthetic compounds but also synthetic products. The library includes compounds featuring fused-ring systems, encompassing a broad spectrum of structures that spans from commonly found ring systems (e.g., pyridine) to innovative bicycles (e.g., pyrazolo[1,5-c]pyrimidine). Such a varied collection of scaffolds underscores the library’s potential for occupying a wide range of the available chemical space. More specifically, the CRAFT fragment library revealed a total of 122 distinct scaffolds, highlighting the rich structural diversity within the library. Among these, the top five scaffolds after benzene are furo[3,2-b]pyridine 1, pyridine 2, indazole 3, imidazo[1,2-a]pyridine 4, and piperazine 5 (Figure 5).

Figure 5.

Figure 5

Open in a new tab

Top five most common scaffolds in the CRAFT fragment library and examples of some novel scaffolds contained in this library.

These scaffolds represent important structural motifs commonly encountered in bioactive molecules.51 Notably, other significant scaffolds included 1H-pyrazolo[3,4-c]pyridine 6 and 2,6-naphthyridine 7 (Figure 5), which exemplifies novel scaffolds synthesized previously by our group.52 Further analysis indicated that only 10 of the 122 scaffolds were nonheterocyclic, while 88 contained nitrogen within their core structure, suggesting a strong emphasis on heterocyclic chemistry. Moreover, 77 of these scaffolds were fused, further supporting the hypothesis that many of them closely resemble natural product structures, which are often characterized by fused-ring systems. This unique composition not only broadens the chemical diversity of the library but also explains why it demonstrates a higher overlap with the COCONUT database compared with other chemical libraries, providing a richer source for exploring bioactive and functional molecular scaffolds.

It is useful to identify the relation between the different databases for the identification of structures conserved in the different databases. However, the diversity in the libraries makes it possible to generate a database of the collections and opens the possibility of their analysis in terms of various chemical and biological implications.53

3.3. Structural Content, Composition, and Complexity of the Compound and Fragment Libraries

As described in the Methodology section, fragments and “Fragment RO3” were analyzed using fourteen constitutional and complexity descriptors. Table 6 describes the constitutional and complexity descriptors of NPs from the LANaPDB and the COCONUT. Tables 7 and 8 summarize constitutional and complexity descriptors from fragments and “Fragment RO3”. The fraction of carbon, oxygen, and nitrogen atoms for LANaPDB’s compounds was 0.8, 0.19, and 0.01, respectively, and close to COCONUT’s compounds 0.78, 0.18, and 0.04, respectively. Chavez-Hernandez et al. (2024) reported the structural composition of NPDBEjeCol-, BIOFACQUIM-, NuBBEDB-, PeruNPDB-, and FDA-approved drugs. Similar to this study’s COCONUT and LANaPDB, all databases of NPs maintain a tendency for a higher fraction of carbon atoms, presenting a higher content of oxygen atoms than nitrogen atoms. While this behavior is distinct in libraries that contain synthetic compounds, for example, as FDA-approved drugs (the fraction of carbon 0.68, oxygen atoms 0.17, and nitrogen 0.09).54 This corresponds to what has been reported in the literature.55,56

Table 7. Constitutional and Complexity Descriptors of NP Fragments and Reference Librariesa.

data set COCONUT LANaPDB CRAFT Enamine ChemDiv Maybridge Life Chemicals
carbon atoms 25.95 24.48 14.90 10.55 12.20 12.77 12.59
oxygen atoms 10.11 8.11 2.22 1.41 2.05 1.86 1.76
nitrogen atoms 0.41 0.22 2.38 2.02 2.20 2.11 2.86
fraction of carbons 0.71 0.75 0.72 0.70 0.70 0.70 0.70
fraction of oxygens 0.27 0.24 0.11 0.10 0.12 0.10 0.10
fraction of nitrogens 0.01 0.01 0.12 0.14 0.13 0.12 0.16
fraction of sp3 carbons 0.64 0.71 0.18 0.42 0.31 0.25 0.41
fraction of chiral carbons 0.29 0.32 0.01 0.04 0.02 0.01 0.03
molecular weight 517.31 460.51 291.65 215.12 249.30 264.73 252.55
heavy atoms 36.55 32.82 20.54 15.07 17.20 18.00 17.86
rings 3.38 3.96 2.79 1.87 2.10 2.15 2.37
aliphatic rings 2.36 3.31 0.61 0.70 0.41 0.36 0.79
aromatic rings 1.01 0.65 2.18 1.17 1.69 1.79 1.58
heterocycles 1.73 1.45 1.50 1.14 1.12 0.99 1.69
aliphatic heterocycles 1.5 1.18 0.51 0.53 0.32 0.24 0.59
aromatic heterocycles 1.01 0.65 2.18 1.17 1.69 1.79 1.58
spiro atoms 0.15 0.75 0.01 0.01 0.01 0.01 0.01
bridgehead atoms 0.51 1.52 0.02 0.02 0.03 0.05 0.06
Open in a new tab
a

Mean value of the distribution.

Table 8. Constitutional and Complexity Descriptors of NPs and Reference Libraries Following RO3a.

data set COCONUT LANaPDB CRAFT Enamine ChemDiv Maybridge Life Chemicals
carbon atoms 9.97 10.04 9.7 10.61 9.44 9.52 11.04
oxygen atoms 1.82 2.19 1.15 1.25 1.16 1.14 1.22
nitrogen atoms 0.64 0.14 1.76 1.77 1.6 1.49 1.9
fraction of carbons 0.78 0.8 0.72 0.72 0.73 0.72 0.73
fraction of oxygens 0.15 0.18 0.09 0.09 0.09 0.09 0.08
fraction of nitrogens 0.06 0.01 0.14 0.12 0.13 0.12 0.13
fraction of sp3 carbons 0.55 0.6 0.16 0.44 0.38 0.28 0.47
fraction of chiral carbons 0.14 0.17 0.01 0.05 0.03 0.02 0.04
molecular weight 177.67 173.01 195.81 210.09 189.9 192.86 213.67
heavy atoms 12.57 12.39 13.34 14.76 12.88 13.04 14.99
rings 1.54 1.51 1.88 1.84 1.61 1.66 2.06
aliphatic rings 1 1.18 0.38 0.78 0.49 0.44 0.95
aromatic rings 0.54 0.33 1.51 1.07 1.12 1.23 1.1
heterocycles 0.73 0.61 1.1 1.06 0.88 0.79 1.32
aliphatic heterocycles 0.54 0.54 0.3 0.59 0.36 0.27 0.69
aromatic heterocycles 0.54 0.33 1.51 1.07 1.12 1.23 1.1
piro atoms 0.05 0.07 0 0.02 0.01 0.01 0.02
bridgehead atoms 0.17 0.14 0 0.02 0.04 0.08 0.11
Open in a new tab
a

Mean value of the distribution.

The fraction of carbon, nitrogen, and oxygen atoms for LANaPDB’s fragments was 0.75, 0.24, and 0.01, respectively, and was similar to that of LANaPDB’s “Fragment RO3” (0.8, 0.18, and 0.01, respectively). However, the total number of carbon atoms shows a difference in the fragments from LANaPDB with respect to the reference libraries. The reference data set contains between 10 and 15 carbon atoms, while the LANaPDB fragments have a mean of 24. Corresponding, the mean values of the molecular-weight LANaPDB are 460.51 and the other reference databases are between 215 and 292. Though, the largest fragments are those obtained from COCONUT with an average value of molecular weight and carbon atoms, 517 and 26, respectively.

A reduction in the fraction of oxygens of reference libraries ca. 0.1 compared to LANaPDB was identified, finding values with a range of 0.10–0.12 for fragments and 0.8–0.9 for “Fragment RO3”. In contrast, the fraction of nitrogen increased >0.1 in reference libraries compared to the displayed value of LANaPDB, with a range of 0.12–0.16 for fragments and 0.12–0.14 for “Fragment RO3”. This is consistent with reports in the literature, the NP often has more oxygen atoms and fewer nitrogen atoms than synthetic molecules. This may be due to the fact that while nitrogen is a necessary component of nature, it is more specialized and therefore less prevalent in organic compounds, while oxygen is a frequent constituent in the biochemical reactions of NP and is so found in more metabolites.55,57

The carbon fractions of the 0.71 COCONUT fragments were similar to “Fragment RO3” of COCONUT 0.78. COCONUT has a similar behavior to that presented by LANaPDB in that the fraction of oxygen decreases and nitrogen increases in reference databases.

Complexity descriptors (Table 6), sp3, and chiral carbon fractions of COCONUT’s compounds were 0.57 and 0.18, respectively, and for LANaPDB’s compounds were 0.59 and 0.21, respectively. These values are similar between NP compounds though LANaPDB compounds were the most complex, presenting the higher mean in both descriptors. The sp3 and chiral carbon fractions of COCONUT’s compounds were similar to the “Fragment RO3” of COCONUT (0.55 and 0.14). A difference was observed with the COCONUT fragments with a higher fraction of sp3, carbons (0.64), and fraction of chiral carbons (0.29). The sp3 (0.6) and chiral carbon (0.17) fractions of LANaPDB compounds are more similar to the values obtained for LANaPDB’s “Fragment RO3” (0.6 and 0.17) and LANaPDB’s fragments (0.71 and 0.32).

We considered the NP data sets to be more complex than other library references. The fraction of sp3 and chiral carbon42 of fragments (sp3 = 0.18–0.41 and chiral = 0.01–0.04) and “Fragment RO3” (sp3 = 0.16–0.47 and chiral = 0.01–0.05) for NP was the highest regarding reference fragments and “Fragment RO3” of Enamine, ChemDiv, Maybridge, and Life Chemicals.

3.4. Structural Similarity

Figure 6 shows the cumulative distribution function and summary statistics of the pairwise Tanimoto similarity using Morgan2 (1024 bit) and MACCS (166 bit) key fingerprints for the fragment libraries. The data sets are distinguished by different colors: COCONUT (cyan), LANaPDB (green), CRAFT (yellow), Enamine (black), ChemDiv (blue), Maybridge (purple), and Life Chemicals (red). Due to the large number of fragments obtained from COCONUT, LANaPDB, ChemDiv, and Life Chemicals (more than 65,248 fragments, Figure 6), ten subsets of 5000 structures were randomly selected from each data set to compute the fingerprint-based similarity. The similarity distribution of the pairwise similarity values computed with the Tanimoto coefficient and the fingerprints MACCS keys and Morgan 2, along with the summary statistics in Figure 5, indicates that Maybridge, ChemDiv, CRAFT, and Enamine were the most diverse fragment libraries, followed by Life Chemicals and the NP-based fragment libraries, COCONUT and LANaPDB. The NP-based fragments were the least diverse, which was unexpected. This could be because of the vast number of fragments and the data variability that are being studied. The distribution of the data is in a considerably wider range and the curve is pushed toward greater similarity compared to the reference libraries.

Figure 6.

Figure 6

Open in a new tab

Cumulative distribution functions of the pairwise Tanimoto similarity using Morgan2 (1024 bit) and MACCS keys (166 bit) of Fragments from COCONUT (cyan), LANaPDB (green), CRAFT (yellow), Enamine (black), ChemDiv (blue), Maybridge (purple), and Life Chemicals (red). The distribution of ten subsets of 5000 fragments randomly selected from COCONUT, LANaPDB, ChemDiv, and Life Chemicals is plotted. The table summarizes the median value of the distributions.

Figure 7 shows the cumulative distribution function and summary statistics of the pairwise Tanimoto similarity using Morgan2 (1024 bit) and MACCS key (166 bit) fingerprints for the “Fragment RO3”. The data sets are color-coded using the same colors as those in Figure 6. The cumulative distribution function and summary statistics indicate that, in general, the COCONUT, Maybridge, ChemDiv, CRAFT, and LANaPDB were the most diverse “Fragment RO3” libraries, followed by Life Chemicals and Enamine. It is interesting to note that the distribution of the NPs is the most diverse and the most similar when looking at the results for the fragments and fragment RO3 (Figures 6 and 7). This suggests that focusing on the “Fragment RO3”, they capture a sizable portion of the diversity of the NPs. Notably, the structural diversity of the fragment libraries, as captured by the fingerprints, is still large and it is not compromised or restricted by the RO3 rules.

Figure 7.

Figure 7

Open in a new tab

Cumulative distribution functions of the pairwise Tanimoto similarity using Morgan2 (1024 bit) and MACCS keys (166 bit) of “Fragment RO3” from COCONUT (cyan), LANaPDB (green), CRAFT (yellow), Enamine (black), ChemDiv (blue), Maybridge (purple), and Life Chemicals (red). The table summarizes the median value of the distributions.

3.5. Synthetic Accessibility

Figure 8 shows the distribution of density for the SA score of compounds in LANaPDB and COCONUT (Figure 8A), fragment libraries (8B), and “Fragment RO3” (8C) for all libraries. The SA score values for compounds in NP databases are similar (for example, the mean SA score value for COCONUT is 4.85 and that for LANaPDB is 4.53). For both NP databases, 86% of the compounds are found with a SA value ≤6, meaning that this fraction of structures is easy to synthesize (according to this approximation).

Figure 8.

Figure 8

Open in a new tab

Synthetic accessibility (SA) score of (A) compounds in COCONUT (cyan) and LANaPDB (green), (B) fragment libraries, and (C) “Fragment RO3”. Data sets are shown in different colors.

The SA score values for LANaPDB and the COCONUT fragment libraries (Figure 8B) (mean values of 5.64 and 4.85, respectively) are larger than the entire compounds in the LANaPDB and the COCONUT libraries. This result could be associated with the structural complexity that is considered in the SA score, being more difficult to synthesize the entire NP due to the presence of macrocycles, fused bicycles, and bridge bicycles.34Figure 8B also indicates that the fragment libraries of Maybridge (mean SA score of 2.39), ChemDiv (2.43), Enamine (2.57), Life Chemicals (2.61), and CRAFT (2.65) had the lowest, most favorable SA scores. Only four molecules of the 1202 CRAFT fragments have an SA score higher than 6. It is expected that the reference fragment libraries would show more favorable SA score values, because the fragments are commercially available from various vendors. The fragments comprising the CRAFT library have been obtained experimentally based on new heterocyclic scaffolds and NP.32 LANaPDB and the COCONUT fragments had higher SA score values (Figure 8B). In fact, around 43% (32,039 structures) of LaNAPDB and 14% (370,800 structures) of COCONUT have SA score values larger than 6, which emphasize the structural complexity and challenge to synthesize NPs.

Overall, all “Fragment RO3” databases (Figure 8C) had a mean SA score in the range considered easy to synthesize (SA value ≤6): Maybridge (2.33), ChemDiv (2.44), Enamine (2.54), CRAFT (2.56), Life Chemicals (2.62), COCONUT (3.41), and LANaPDB (3.53). Only 1% of the “Fragment RO3” databases derived from COCONUT and LANaPDB had SA score values larger than 6. Taken together, it can be summarized that in general and as expected, “Fragments RO3” are more synthetically feasible than fragments not following RO3.

3.6. Visual Representation of the Chemical Space and Chemical Multiverse of Fragment RO3

Figure 9 shows a visual representation of the chemical space of NPs and commercial “Fragment RO3” libraries generated with TMAP using MACCS keys. Figure 10 depicts the chemical space of the same libraries with TMAP based on Morgan2. The visualization with Morgan2 is very similar to the one generated with Morgan3 fingerprints (Figure S5 in the Supporting Information). Figures 9 and 10 libraries are represented in different colors: cyan (COCONUT), yellow (CRAFT), green (LANaPDB), black (Enamine), red (Life Chemicals), and blue (ChemDiv). The visualization of the chemical multiverse (chemical space with different representations, Figures 9 and 10) indicates that “Fragments RO3” from COCONUT cover a large region of the space followed by ChemDiv, Life Chemicals, Enamine, LANaPDB, and CRAFT. This is in line with the quantitative diversity analysis that shows that CRAFT was the fourth most diverse “Fragment RO3” library according to MACCS keys and Morgan2 (Figure 7).

Figure 9.

Figure 9

Open in a new tab

Chemical space visualization of NPs and commercial “Fragment RO3” using TMAP and MACCS keys (166 bit). Data sets are shown in different colors, as indicated in the legend. Chemical space of “Fragment RO3” was split into six panels: (A) All “Fragment RO3”; (B) COCONUT, LANaPDB, and CRAFT; (C) COCONUT and Maybridge; (D) COCONUT and ChemDiv; (E) COCONUT and Enamine; and (F) COCONUT and CRAFT.

Figure 10.

Figure 10

Open in a new tab

Chemical space visualization of NPs and commercial “Fragment RO3” using TMAP and Morgan2 (1024 bit). Data sets are shown in different colors, as indicated in the legend. Chemical space of “Fragment RO3” was split into six panels: (A) All “Fragment RO3”; (B) COCONUT, LANaPDB, and CRAFT; (C) COCONUT and Maybridge; (D) COCONUT and ChemDiv; (E) COCONUT and Enamine; and (F) COCONUT and CRAFT.

Figures S6, S7, and S8 show a visual representation of the chemical space of the “Fragment RO3” libraries using t-SNE, MACCS keys, and Morgan2 and Morgan3 (Figures S6, S7, and S8 in the Supporting Information). The visualization with Morgan2 (Figure S7) is very similar to the one generated with Morgan3 fingerprints (Figure S8). In agreement with the TMAPs (Figure 8), “Fragments RO3” of COCONUT cover a large surface of the chemical space generated with MACCS keys (Figure S6), Morgan2 (Figure S7), and Morgan3 (Figure S8). Figures 9B–F and 10B–F indicate that “Fragments RO3” of LANaPDB, CRAFT, Maybridge, ChemDiv, Enamine, and Life Chemicals share the same chemical space as “Fragments RO3” of COCONUT. However, “Fragments RO3” of LANaPDB and CRAFT share chemical space regions different from those of “Fragments RO3” of Maybridge, ChemDiv, Enamine, and Life Chemicals. These results mean that although the CRAFT fragments are synthetic and NP inspired,32 they retain similar structural features from NP fragments and are to be expected to overlap with the chemical space of NP fragments like COCONUT and LANaPDB. In general, t-SNE and TMAP had similar results. Still, the advantage of TMAP48 in contrast with t-SNE is preserving all information about the chemical structures of fragments as possible. t-SNE is a nonlinear dimension reduction method,49 and more information is lost based on the number of descriptors used.

These results underscore the significant structural diversity offered by natural products, suggesting their potential as a rich source for the discovery of novel compounds in drug development. NPs have historically played a crucial role in the development of many therapeutic agents, accounting for a substantial percentage of drugs currently in use.58 This diversity not only provides a vast array of unique scaffolds but also facilitates the identification of compounds with distinct biological activities.59 Furthermore, recent advances in techniques such as high-throughput screening and cheminformatics have enhanced our ability to explore the vast chemical space represented by natural products, leading to the identification of promising candidates for various therapeutic targets and different neglected and emerging diseases.60−63

4. Conclusions

Herein, we analyzed the contents, properties, and chemical diversity of fragment libraries obtained from the latest releases of the COCONUT and LANaPDB, CRAFT, and commercial fragment libraries. It was concluded that NPs have the highest percentage of unique fragments and “Fragment RO3”. These results highlight the great structural diversity provided by the NPs. Also, NP fragments had higher values of sp3 and fraction of chiral carbons than the reference fragment libraries ChemDiv, Maybridge, Enamine, and Life Chemicals. Similarly, CRAFT fragments and “Fragment RO3” had the highest percentage of unique fragments as compared to the fragment libraries derived from COCONUT and LANaPDB. It was also found that the fragments from NPs (including the one that complies with the RO3) had a similar number of heterocycles and carbons and similar values of nitrogens, oxygens, sp3, and chiral carbon fraction, meaning that NP fragments retained structural features and complexity features of NPs from which they are derived. We also found that “Fragments RO3” from COCONUT were the most structurally diverse (as quantified using molecular fingerprints), meaning that the filters of the RO3 reduced the total number of COCONUT fragments but the COCONUT fragments that comply with the RO3 have a high diversity. All fragment libraries herein obtained and curated are freely available at https://github.com/DIFACQUIM/Fragment-libraries-from-large-synthetic-compounds-and-natural-products-collections.git.

Acknowledgments

V. R.-C. and A.L.C.-H. are thankful to CONAHCyT, Mexico, for the scholarship number 2053818 and 847870. Authors thank the Direction General de Cómputo y de Tecnologías de la Información y Comunicación (DGTIC), UNAM, for the computational resources to use Miztli supercomputer at UNAM under project LANCAD-UNAM-DGTIC-335. C.H.A. thanks the National Council for Scientific and Technological Development (CNPq grants #483659/2013-4, #444750/2024-0), BRICS STI COVID-19 (grant #441038/2020-4), and Goias State Research Foundation (FAPEG grant #202010267000272). C.H.A. and F.S.E. are CNPq research fellows. F.S.E. and T.O.A.P are supported by the National Institute of Health grant (5R01AI160379-03). F.S.E. thanks the National Council for Scientific and Technological Development (CNPq grant #443750/2023-8). R.M.N., K.M.A., and W.J.L.S. are supported by fellowships from Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—Brasil (CAPES)—Finance Code 001.

Glossary

Abbreviations

BRICS

Breaking of Retrosynthetically Interesting Chemical Substructures

CRAFT

Center for Research and Advancements in Fragments and Molecular Targets

COCONUT

Collection of Open Natural Products

FBDD

fragment-based drug discovery

HBAs

hydrogen-bond acceptors

HBDs

hydrogen-bond donors

LANaPDB

Latin America Natural Product Database

Log P

partition coefficient octanol/water

NPs

natural products

MACCS

Molecular ACCes System

MORTAR

MOlecule fRagmenTAtion fRamework

MW

molecular weight

RECAP

Retrosynthetic Combinatorial Analysis Procedure

RB

rotatable bonds

SA

synthetic accessibility

SMILES

Simplified Molecular Input Line Entry System

TMAP

Tree MAP

TPSA

topological polar surface area.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c01420.

  • Unique and common fragments between CRAFT, LANaPDB, COCONUT, and reference libraries; cumulative distribution functions of the pairwise Tanimoto similarity using Morgan3 of Fragments and “Fragments RO3” from COCONUT, LANaPDB, CRAFT, Enamine, ChemDiv, Maybridge, and Life Chemicals; and chemical space visualization of commercial “Fragments RO3” using TMAP and t-SNE with Morgan3 and MACCS keys (PDF)

The authors declare no competing financial interest.

Special Issue

Published as part of ACS Omegaspecial issue “Chemistry in Brazil: Advancing through Open Science”.

Supplementary Material

ao5c01420_si_001.pdf (1.9MB, pdf)

References

  1. Chávez-Hernández A. L.; Sánchez-Cruz N.; Medina-Franco J. L. Fragment Library of Natural Products and Compound Databases for Drug Discovery. Biomolecules 2020, 10, 1518. 10.3390/biom10111518. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Thomford N. E.; Senthebane D. A.; Rowe A.; Munro D.; Seele P.; Maroyi A.; Dzobo K. Natural Products for Drug Discovery in the 21st Century: Innovations for Novel Drug Discovery. Int. J. Mol. Sci. 2018, 19, 1578. 10.3390/ijms19061578. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Zhao J.-X.; Yue J.-M. Frontier Studies on Natural Products: Moving toward Paradigm Shifts. Sci. China Chem. 2023, 66, 928–942. 10.1007/s11426-022-1512-0. [DOI] [Google Scholar]
  4. Over B.; Wetzel S.; Grütter C.; Nakai Y.; Renner S.; Rauh D.; Waldmann H. Natural-Product-Derived Fragments for Fragment-Based Ligand Discovery. Nat. Chem. 2013, 5, 21–28. 10.1038/nchem.1506. [DOI] [PubMed] [Google Scholar]
  5. Pahl A.; Waldmann H.; Kumar K. Exploring Natural Product Fragments for Drug and Probe Discovery. Chimia 2017, 71, 653–660. 10.2533/chimia.2017.653. [DOI] [PubMed] [Google Scholar]
  6. Liu M.; Quinn R. J. Fragment-Based Screening with Natural Products for Novel Anti-Parasitic Disease Drug Discovery. Expert Opin. Drug Discov. 2019, 14, 1283–1295. 10.1080/17460441.2019.1653849. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Karageorgis G.; Foley D. J.; Laraia L.; Brakmann S.; Waldmann H. Pseudo Natural Products-Chemical Evolution of Natural Product Structure. Angew. Chem., Int. Ed. Engl. 2021, 60, 15705–15723. 10.1002/anie.202016575. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Grigalunas M.; Brakmann S.; Waldmann H. Chemical Evolution of Natural Product Structure. J. Am. Chem. Soc. 2022, 144, 3314–3329. 10.1021/jacs.1c11270. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Sorokina M.; Merseburger P.; Rajan K.; Yirik M. A.; Steinbeck C. COCONUT Online: Collection of Open Natural Products Database. J. Cheminf. 2021, 13, 2. 10.1186/s13321-020-00478-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Chandrasekhar V.; Rajan K.; Kanakam S. R. S.; Sharma N.; Weißenborn V.; Schaub J.; Steinbeck C. COCONUT 2.0: A Comprehensive Overhaul and Curation of the Collection of Open Natural Products Database. Nucleic Acids Res. 2025, 53 (D1), D634–D643. 10.1093/nar/gkae1063. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Gómez-García A.; Medina-Franco J. L. Progress and Impact of Latin American Natural Product Databases. Biomolecules 2022, 12, 1202. 10.3390/biom12091202. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Gómez-García A.; Jiménez D. A. A.; Zamora W. J.; Barazorda-Ccahuana H. L.; Chávez-Fumagalli M. A. ´.; Valli M.; Andricopulo A. D.; Bolzani V. da S.; Olmedo D. A.; Solís P. N.; Núñez M. J.; Rodríguez Pérez J. R.; Valencia Sánchez H. A.; Cortés Hernández H. F.; Medina-Franco J. L. Navigating the Chemical Space and Chemical Multiverse of a Unified Latin American Natural Product Database: LANaPDB. Pharmaceuticals 2023, 16, 1388. 10.3390/ph16101388. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Gómez-García A.; Prinz A.-K.; Jiménez D. A. A.; Zamora W. J.; Barazorda-Ccahuana H. L.; Chávez-Fumagalli M. Á.; Valli M.; Andricopulo A. D.; da S Bolzani V.; Olmedo D. A.; Solís P. N.; Núñez M. J.; Rodríguez Pérez J. R.; Sánchez H. A. V.; Cortés Hernández H. F.; Mosquera Martinez O. M.; Koch O.; Medina-Franco J. L. Updating and Profiling the Natural Product-Likeness of Latin American Compound Libraries. Mol. Inform. 2024, 43, e202400052 10.1002/minf.202400052. [DOI] [PubMed] [Google Scholar]
  14. Bissaro M.; Sturlese M.; Moro S. The Rise of Molecular Simulations in Fragment-Based Drug Design (FBDD): An Overview. Drug Discov. Today 2020, 25, 1693–1701. 10.1016/j.drudis.2020.06.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Khedkar N. R.; Sindkhedkar M.; Joseph A. Fragment-Based Drug Discovery: Small Fragments, Big Impact - Success Stories of Approved Oncology Therapeutics. Bioorg. Chem. 2025, 156, 108197. 10.1016/j.bioorg.2025.108197. [DOI] [PubMed] [Google Scholar]
  16. Congreve M.; Carr R.; Murray C.; Jhoti H. A “Rule of Three” for Fragment-Based Lead Discovery?. Drug Discov. Today 2003, 8, 876–877. 10.1016/S1359-6446(03)02831-9. [DOI] [PubMed] [Google Scholar]
  17. Bon M.; Bilsland A.; Bower J.; McAulay K. Fragment-Based Drug Discovery-the Importance of High-Quality Molecule Libraries. Mol. Oncol. 2022, 16, 3761–3777. 10.1002/1878-0261.13277. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Wang L.; Gao J.; Ma R.; Liu Y.; Liu M.; Zhong F.; Hu J.; Li S.; Wu J.; Jiang H.; Zhang J.; Ruan K. Recent Progress in Fragment-Based Drug Discovery Facilitated by NMR Spectroscopy. Magn. Reson. Lett. 2022, 2, 107–118. 10.1016/j.mrl.2021.100025. [DOI] [Google Scholar]
  19. Velvadapu V.; Farmer B. T.; Reitz A. B.. Fragment-Based Drug Discovery. In The Practice of Medicinal Chemistry; Elsevier, 2015; pp 161–180. [Google Scholar]
  20. Osborne J.; Panova S.; Rapti M.; Urushima T.; Jhoti H. Fragments: Where Are We Now?. Biochem. Soc. Trans. 2020, 48, 271–280. 10.1042/BST20190694. [DOI] [PubMed] [Google Scholar]
  21. Li Q. Application of Fragment-Based Drug Discovery to Versatile Targets. Front. Mol. Biosci. 2020, 7, 180. 10.3389/fmolb.2020.00180. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Tang Y.; Moretti R.; Meiler J. Recent Advances in Automated Structure-Based De Novo Drug Design. J. Chem. Inf. Model. 2024, 64, 1794–1805. 10.1021/acs.jcim.4c00247. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Lewell X. Q.; Judd D. B.; Watson S. P.; Hann M. M. RECAP--Retrosynthetic Combinatorial Analysis Procedure: A Powerful New Technique for Identifying Privileged Molecular Fragments with Useful Applications in Combinatorial Chemistry. J. Chem. Inf. Comput. Sci. 1998, 38, 511–522. 10.1021/ci970429i. [DOI] [PubMed] [Google Scholar]
  24. Degen J.; Wegscheid-Gerlach C.; Zaliani A.; Rarey M. On the Art of Compiling and Using “Drug-like” Chemical Fragment Spaces. ChemMedChem 2008, 3, 1503–1507. 10.1002/cmdc.200800178. [DOI] [PubMed] [Google Scholar]
  25. Bänsch F.; Schaub J.; Sevindik B.; Behr S.; Zander J.; Steinbeck C.; Zielesny A. MORTAR: A Rich Client Application for in Silico Molecule Fragmentation. J. Cheminf. 2023, 15, 1. 10.1186/s13321-022-00674-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Fritsch S.; Neumann S.; Schaub J.; Steinbeck C.; Zielesny A. ErtlFunctionalGroupsFinder: Automated Rule-Based Functional Group Detection with the Chemistry Development Kit (CDK). J. Cheminf. 2019, 11, 37. 10.1186/s13321-019-0361-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Schaub J.; Zander J.; Zielesny A.; Steinbeck C. Scaffold Generator: A Java Library Implementing Molecular Scaffold Functionalities in the Chemistry Development Kit (CDK). J. Cheminf. 2022, 14, 79. 10.1186/s13321-022-00656-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Fragment Collection. https://enamine.net/compound-collections/fragment-collection (accessed June 11, 2024).
  29. ChemDiv INC - Fully integrated target to clinic contract research organization. https://www.chemdiv.com (accessed June 11, 2024).
  30. Screening M. H.-T.Increasing success rates in your drug discovery research. https://www.thermofisher.in/content/dam/fishersci/chemicals/pdf/Maybridge-Drugs-discovery-brochure.pdf (accessed June 11, 2024).
  31. Fragment Libraries. https://lifechemicals.com/screening-libraries/fragment-libraries (accessed June 11, 2024).
  32. Andrade C. H.; Nonato M. C.; da Silva Emery F. Introducing CRAFT: The Center for Research and Advancement in Fragments and Molecular Targets. ACS Med. Chem. Lett. 2024, 15, 1174–1177. 10.1021/acsmedchemlett.4c00296. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Silva D. G.; Emery F. d. S. Strategies Towards Expansion of Chemical Space of Natural Product-Based Compounds to Enable Drug Discovery. Braz. J. Pharm. Sci. 2018, 54, 16. 10.1590/s2175-97902018000001004. [DOI] [Google Scholar]
  34. Ertl P.; Schuffenhauer A. Estimation of Synthetic Accessibility Score of Drug-like Molecules Based on Molecular Complexity and Fragment Contributions. J. Cheminf. 2009, 1, 8. 10.1186/1758-2946-1-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Schneider G.; Fechner U. Computer-Based de Novo Design of Drug-like Molecules. Nat. Rev. Drug Discovery 2005, 4, 649–663. 10.1038/nrd1799. [DOI] [PubMed] [Google Scholar]
  36. Gómez-García A.; Acuña Jiménez D. A.; Zamora W. J.; Barazorda-Ccahuana H. L.; Chávez-Fumagalli M. A. ´.; Valli M.; Andricopulo A. D.; Bolzani V. da S.; Olmedo D. A.; Solís P. N.; Núñez M. J.; Rodríguez Pérez J. R.; Valencia Sánchez H. A.; Cortés Hernández H. F.; Mosquera Martinez O. M.; Medina-Franco J. L. Latin American Natural Product Database (LANaPDB): An Update. J. Chem. Inf. Model. 2024, 64, 8495–8509. 10.1021/acs.jcim.4c01560. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Weininger D.; Weininger A.; Weininger J. L. SMILES. 2. Algorithm for Generation of Unique SMILES Notation. J. Chem. Inf. Comput. Sci. 1989, 29, 97–101. 10.1021/ci00062a008. [DOI] [Google Scholar]
  38. Landrum G.RDKit. https://www.rdkit.org (accessed June 11, 2024).
  39. MolVS : Molecule Validation and Standardization — MolVS 0.1.1 documentation. https://molvs.readthedocs.io/en/latest/ (accessed June 11, 2024).
  40. Sánchez-Cruz N.; Pilón-Jiménez B. A.; Medina-Franco J. L. Functional Group and Diversity Analysis of BIOFACQUIM: A Mexican Natural Product Database. F1000Res 2019, 8, 2071. 10.12688/f1000research.21540.1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Feher M.; Schmidt J. M. Property Distributions: Differences between Drugs, Natural Products, and Molecules from Combinatorial Chemistry. J. Chem. Inf. Comput. Sci. 2003, 43, 218–227. 10.1021/ci0200467. [DOI] [PubMed] [Google Scholar]
  42. Krzyzanowski A.; Pahl A.; Grigalunas M.; Waldmann H. Spacial Score–A Comprehensive Topological Indicator for Small-Molecule Complexity. J. Med. Chem. 2023, 66, 12739–12750. 10.1021/acs.jmedchem.3c00689. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Jaccard P. Etude Comparative de La Distribution Florale Dans Une Portion Des Alpes et Des Jura. Bull. Soc. Vaud. sci. nat. 1901, 37, 547–579. 10.5169/seals-266450. [DOI] [Google Scholar]
  44. Durant J. L.; Leland B. A.; Henry D. R.; Nourse J. G. Reoptimization of MDL Keys for Use in Drug Discovery. J. Chem. Inf. Comput. Sci. 2002, 42, 1273–1280. 10.1021/ci010132r. [DOI] [PubMed] [Google Scholar]
  45. Rogers D.; Hahn M. Extended-Connectivity Fingerprints. J. Chem. Inf. Model. 2010, 50, 742–754. 10.1021/ci100050t. [DOI] [PubMed] [Google Scholar]
  46. Virshup A. M.; Contreras-García J.; Wipf P.; Yang W.; Beratan D. N. Stochastic Voyages into Uncharted Chemical Space Produce a Representative Library of All Possible Drug-like Compounds. J. Am. Chem. Soc. 2013, 135, 7296–7303. 10.1021/ja401184g. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Medina-Franco J. L.; Chávez-Hernández A. L.; López-López E.; Saldívar-González F. I. Chemical Multiverse: An Expanded View of Chemical Space. Mol. Inform. 2022, 41, e2200116 10.1002/minf.202200116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Probst D.; Reymond J.-L. Visualization of Very Large High-Dimensional Data Sets as Minimum Spanning Trees. J. Cheminf. 2020, 12, 12. 10.1186/s13321-020-0416-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Van der Maaten L.; Hinton G. Visualizing Data Using t-SNE. J. Mach. Learn. Res. 2008, 9, 2579–2605. [Google Scholar]
  50. Marvin 17.21.0. ChemAxon. https://www.chemaxon.com (accessed June 11, 2024).
  51. Shearer J.; Castro J. L.; Lawson A. D. G.; MacCoss M.; Taylor R. D. Rings in Clinical Trials and Drugs: Present and Future. J. Med. Chem. 2022, 65, 8699–8712. 10.1021/acs.jmedchem.2c00473. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Silva Júnior P. E.; Rezende L. C. D.; Gimenes J. P.; Maltarollo V. G.; Dale J.; Trossini G. H. G.; Emery F. S.; Ganesan A. Synthesis of Two “Heteroaromatic Rings of the Future” for Applications in Medicinal Chemistry. RSC Adv. 2016, 6, 22777–22780. 10.1039/C6RA01099G. [DOI] [Google Scholar]
  53. Grigalunas M.; Burhop A.; Zinken S.; Pahl A.; Gally J.-M.; Wild N.; Mantel Y.; Sievers S.; Foley D. J.; Scheel R.; Strohmann C.; Antonchick A. P.; Waldmann H. Natural Product Fragment Combination to Performance-Diverse Pseudo-Natural Products. Nat. Commun. 2021, 12, 1883. 10.1038/s41467-021-22174-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Chávez-Hernández A. L.; Rodríguez-Pérez J. R.; Cortés-Hernández H. F.; Valencia-Sanchez H. A.; Chávez-Fumagalli M. A.´; Medina-Franco J. L. Fragment Library of Colombian Natural Products: Generation and Comparative Chemoinformatic Analysis. Drugs and Drug Candidates 2024, 3, 736–750. 10.3390/ddc3040042. [DOI] [Google Scholar]
  55. Chen Y.; Stork C.; Hirte S.; Kirchmair J. NP-Scout: Machine Learning Approach for the Quantification and Visualization of the Natural Product-Likeness of Small Molecules. Biomolecules 2019, 9, 43. 10.3390/biom9020043. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Atanasov A. G.; Zotchev S. B.; Dirsch V. M.; Orhan I. E.; Banach M.; Rollinger J. M.; Barreca D.; Weckwerth W.; Bauer R.; Bayer E. A.; et al. Natural Products in Drug Discovery: Advances and Opportunities. Nat. Rev. Drug Discovery 2021, 20, 200–216. 10.1038/s41573-020-00114-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Chen Y.; Rosenkranz C.; Hirte S.; Kirchmair J. Ring Systems in Natural Products: Structural Diversity, Physicochemical Properties, and Coverage by Synthetic Compounds. Nat. Prod. Rep. 2022, 39, 1544–1556. 10.1039/D2NP00001F. [DOI] [PubMed] [Google Scholar]
  58. Newman D. J.; Cragg G. M. Natural Products as Sources of New Drugs over the Nearly Four Decades from 01/1981 to 09/2019. J. Nat. Prod. 2020, 83, 770–803. 10.1021/acs.jnatprod.9b01285. [DOI] [PubMed] [Google Scholar]
  59. Harvey A. L. Natural Products in Drug Discovery. Drug Discov. Today 2008, 13, 894–901. 10.1016/j.drudis.2008.07.004. [DOI] [PubMed] [Google Scholar]
  60. Mottin M.; Caesar L. K.; Brodsky D.; Mesquita N. C. M. R.; de Oliveira K. Z.; Noske G. D.; Sousa B. K. P.; Ramos P. R. P. S.; Jarmer H.; Loh B.; Zorn K. M.; Foil D. H.; Torres P. M.; Guido R. V. C.; Oliva G.; Scholle F.; Ekins S.; Cech N. B.; Andrade C. H.; Laster S. M. Chalcones from Angelica Keiskei (ashitaba) Inhibit Key Zika Virus Replication Proteins. Bioorg. Chem. 2022, 120, 105649. 10.1016/j.bioorg.2022.105649. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Moreira-Filho J. T.; Neves B. J.; Cajas R. A.; Moraes J. de; Andrade C. H. Artificial Intelligence-Guided Approach for Efficient Virtual Screening of Hits against Schistosoma Mansoni. Future Med. Chem. 2023, 15, 2033–2050. 10.4155/fmc-2023-0152. [DOI] [PubMed] [Google Scholar]
  62. Purificação A. D.; Silva-Mendonça S.; Cruz L. V.; Sacramento C. Q.; Temerozo J. R.; Fintelman-Rodrigues N.; de Freitas C. S.; Godoi B. F.; Vaidergorn M. M.; Leite J. A.; Salazar Alvarez L. C.; Freitas M. V.; Silvac M. F. B.; Martin B. A.; Lopez R. F. V.; Neves B. J.; Costa F. T. M.; Souza T. M. L.; da Silva Emery F.; Andrade C. H.; Nonato M. C. Unveiling the Antiviral Capabilities of Targeting Human Dihydroorotate Dehydrogenase against SARS-CoV-2. ACS Omega 2024, 9, 11418–11430. 10.1021/acsomega.3c07845. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. de Souza Neto L. R.; Montoya B. O.; Brandão-Neto J.; Verma A.; Bowyer S.; Moreira-Filho J. T.; Dantas R. F.; Neves B. J.; Andrade C. H.; von Delft F.; Owens R. J.; Furnham N.; Silva-Jr F. P. Fragment Library Screening by X-Ray Crystallography and Binding Site Analysis on Thioredoxin Glutathione Reductase of Schistosoma Mansoni. Sci. Rep. 2024, 14, 1582. 10.1038/s41598-024-52018-2. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

ao5c01420_si_001.pdf (1.9MB, pdf)

Articles from ACS Omega are provided here courtesy of American Chemical Society

RESOURCES