MACC combination

Four classes of 2D fingerprint types can be distinguished: (i) dictionary-based, (ii) topological or path-based, (iii) circular fingerprints and (iv) pharmacophores. In addition, fingerprints can differ in the atom types or feature classes used or the length of the bit string. In this study, 14 fingerprints belonging to three of the four classes were compared.
The public Molecular ACCess System (MACCS) structural keys
[43 ] are 166 predefined substructures defined as SMARTS and belong to the dictionary-based fingerprint class. They were originally designed for substructure search and typically show a low performance level for virtual screening, thus they are often used as baseline fingerprint for benchmarking studies.
Topological or path-based fingerprints describe combinations of atom types and paths between atom types. In atom pair (AP) fingerprints
[44 (link)], pairs of atoms together with the number of bonds separating them are encoded. In topological torsions (TT)
[37 (link)], on the other hand, four atoms forming a torsion are described. In both AP and TT fingerprints the atom type consists of the element, the number of heavy-atom neighbours and the number of π-electrons.
The RDKit fingerprint, a relative of the well-known Daylight fingerprint
[45 ], is another topological descriptor. Atom-types, the atomic number and aromaticity state, are combined with bond types to hash all branched and linear molecular subgraphs up to a particular size
[42 ]. In this study, a maximum path length of five (RDK5) was used.
Similar to the Daylight fingerprints, certain paths and feature classes of the molecular graph are enumerated and hashed in the Avalon fingerprint
[46 (link)]. There are 16 feature classes which were optimized for substructure search. A detailed description of the feature classes is given in Table
1 and the supplementary material of
[46 (link)].
Circular fingerprints were developed more recently
[47 (link)], and encode circular atom environments up to a certain bond radius from the central atom. If atom types consisting of the element, the number of heavy-atom neighbours, the number of hydrogens, the isotope and ring information are used these fingerprints are called extended-connectivity (EC) fingerprints. Alternatively, pharmacophoric features can be used, yielding functional connectivity (FC) fingerprints. We consider two representations of the fingerprints, bit strings (FP) and count vectors (FC). This gives four types of circular fingerprints: extended-connectivity bit string (ECFP), extended-connectivity count vector (ECFC), feature-connectivity bit string (FCFP) and feature-connectivity count vector (FCFC). The maximum bond length or diameter is added at the end to the name. In this study, the four types of circular fingerprints with a diameter 4, i.e. ECFP4, ECFC4, FCFP4 and FCFC4, as well as ECFP6 were compared. In addition, ECFC0, which is a kind of atom count, was used as a second baseline fingerprint.
For all bit-string fingerprints, a size of 1024 bits was used. However, Sastry et al. found that such a small bit space may result in many collisions which can affect VS performance
[21 (link)]. To investigate this effect a larger bit space, 16384 bits, was used for three fingerprints: long ECFP4 (lECFP4), long ECFP6 (lECFP6) and long Avalon (lAvalon).
All fingerprints were calculated using the RDKit.

The molecular fingerprints used were taken from the benchmarking platform described by Riniker and Landrum [9 (link)] and are listed in Table 1. Although their study focused on results for 14 fingerprints, the associated code [24 ] includes a further 14, mainly additional variants of circular fingerprints but also hashed forms of atom pairs (HashAP) and topological torsions (HashTT). In this study we have used the full set of 28 fingerprints as implemented in the RDKit version 2015.09.2 [25 ].
The fingerprints may be classified as follows. Additional details are in the publication by Riniker and Landrum:

Path-based fingerprints RDKx where x is 5, 6, 7 (hashed branched and linear subgraphs up to size x), TT (topological torsion [26 (link)], a count vector) and a binary vector form HashTT, AP [27 (link)] (atom pair, a count vector) and a binary vector form HashAP.

Substructure keys Avalon [28 (link)], MACCS.

Circular fingerprints The extended-connectivity fingerprints [29 (link)] ECFPx where x is 0, 2, 4, 6, and the corresponding count vectors denoted as ECFCx. Also the feature-class fingerprints FCFPx and corresponding count vectors FCFCx where x is 2, 4, 6.

A length of 1024 bits was used for all binary fingerprints listed above, but for comparison a longer length of 16384 bits was used for a number of fingerprints (as in the original study). This longer version is indicated by the prefix “L”: LAvalon, LECFP6, LECFP4, LFCFP6 and LFCFP4. The Tanimoto coefficient was used to measure similarity for all binary fingerprints, while the Dice coefficient was used for count vectors.

MACC scores were computed as described previously (37 (link)). Briefly, read frequencies were computed in non-overlapping bins of selected size (300 bp) for each titration point independently, normalized to library sizes, and fit with a linear regression. The estimated regression coefficients were corrected to remove dependence on GC content. The corrected values were used as MACC scores.

The similarity degree of the ligands reported in the curated ChEMBL and DrugBank datasets were evaluated as follows. First, an all-against-all 2D-fingerprint-based similarity assessment was performed by using an in-house-developed Python script. The degree of 2D similarity was calculated according to the MACCS and circular (i.e., ECFP4) type of fingerprints available on OpenEye Python toolkits [194 ], and by using the Tanimoto coefficient (Tc) as an index for ligands similarity measurement. Then, the results of the 2D similarity calculations were filtered to retain only those with MACCS and ECFP4 Tanimoto coefficients equal to or higher than 0.8 and 0.3, respectively, in agreement with previously reported studies [196 (link)]. Afterward, extensive 3D shape and atom type similarity calculations of compounds resulting as significantly similar in the 2D estimations were performed by using the ROCS software [117 ]. In this case, the similarity degree between ligands was assessed according to the Tanimoto Combo (TCc) coefficient [117 ]. Finally, the results of the 3D-shape-based similarity calculations were filtered to retain only similarity records with a TCc ≥ 1.5, which is a commonly accepted threshold of similarity [196 (link)].

The benthic diatom N. shiloi was previously identified, and the details are given by Demirel, et al. [74 (link)]. In that study, benthic diatoms were separated from the Aegean Sea in Türkiye and identified according to their morphological characteristics on the basis of observations in bright field and scanning electron microscopy. For DNA extraction, the cultured strain was harvested by centrifugation, and the cell pellet was used in the DNA Kits (Zymo Research) and stored at −20 °C. PCR primers targeted to link 18S rDNA (F, 5′-YACCTGGTTGATCCTGCCAGTAG-3′ and R, 5′-GCTTGATCCTTCTGCAGGTTCACC-3′). PCR protocol was applied, and products were viewed in an agarose gel stained by Jel Safe Dye and checked the gel under a UV light. In the DNA sequencing step, dye-terminator sequencing was performed using the primers, and the nucleotide chromatograms were determined by DNA sequences. The sequences obtained in the mentioned study were deposited in the NCBI GenBank, and the accession numbers of KR149459. Nanofrustulum shiloi (EGEMACC 47) was deposited in Ege University, Microalgae Culture Collection, Izmir, Turkey (EGEMACC—https://ege-macc.ege.edu.tr/ (accessed on 15 February 2023)).