Molecular flexibility is a commonly used, but not easily quantified term. It is at the core of understanding composition and size of a conformational ensemble and contributes to many molecular properties. For many computational workflows, it is necessary to reduce a conformational ensemble to meaningful representatives, however defining them and guaranteeing the ensemble's completeness is difficult. We introduce the concepts of torsion angular bin strings (TABS) as a discrete vector representation of a conformer's dihedral angles and the number of possible TABS (nTABS) as an estimation for the ensemble size of a molecule, respectively. Here, we show that nTABS corresponds to an upper limit for the size of the conformational space of small molecules and compare the classification of conformer ensembles by TABS with classifications by RMSD. Overcoming known drawbacks like the molecular size dependency and threshold picking of the RMSD measure, TABS is shown to meaningfully discretize the conformational space and hence allows e.g. for fast checks of the coverage of the conformational space. The current proof-of-concept implementation is based on the ETKDGv3 conformer generator as implemented in the RDKit and known torsion preferences extracted from small-molecule crystallographic data. Following this logic, the size of the conformational space of a molecule, and thus its molecular flexibility, can be estimated by generating a large ensemble of conformers and then pruning them based on a chosen distance metric. One of the most frequently used descriptors to quantify molecular flexibility is the number of rotatable bonds. Though common, this descriptor suffers from a number of drawbacks. Perhaps the largest of these is that it requires a clear, and ideally easily computed, definition of which bonds are rotatable. There are many of such definitions, e.g., the one proposed by Bath et al., 10 but not one used by all. Furthermore, the number of rotatable bonds, being constrained to integer values, provides a very coarse-grained view of the size of the conformational space and ignores the fact that different types of bonds have different degrees of rotational freedom. To overcome these challenges, Kier developed the ϕ index, 3 which provides a continuous description of the flexibility space derived solely based upon information from the molecular graph. A TABS itself is a vector representation for a conformer reduced to a description of its dihedral angles: Each vector element corresponds to the binned value of the torsion about one rotatable bond in the molecule. The TABS representation discretizes the torsion space and is a form of dimensionality reduction that simplifies the analysis and understanding of conformational ensembles. After this definition, it is also clear that TABS and a torsion fingerprint (TF) 6 are inherently different as the TFs operate on a continuous space, as well as treating ring contributions as average sums. TFD 6 itself and TABS are only comparable if a distance metric between two TABS was defined, which has not been done as part of this initial method development. Theory Common Flexibility Metrics Before describing our new flexibility metric nTABS, we provide a short overview over two of the most commonly used flexibility metrics: number of rotatable bonds and Kier ϕ index. 3 Rotatable-Bond Count The most common definition of a rotatable bond is a single bond that is not part of a ring connecting two atoms, which each have at least one other nonterminal substituent. 14 Refinements typically include aspects like ignoring bonds where one atom has only symmetry-equivalent substituents or including bonds in macrocycles. Here, we use the default rotatable bond definition in the RDKit, 15 described in detail in the Supporting Information S1 . Kier ϕ Index As with the rotatable-bond count, Kier treats flexibility as a structural attribute that can be derived directly from the molecular graph. 3 Kier’s reference point for a perfectly flexible molecule is the infinite chain of carbon atoms with sp 3 hybridization (Csp 3 ), which marks the point where the flexibility index ϕ is defined to be infinite. Figure 1 Example of TABS assignment for two example conformers of a molecule. Regular Torsions In a molecule, each rotatable bond, as identified by a chosen definition, can be associated with a distinct torsion profile. These torsion profiles are influenced by the molecule’s overall structure, giving each dihedral an individual profile. Though each torsion is, in principle, unique, it is possible to assign them to a comparatively small number of archetypes, as for instance introduced by Schärfer et al. 18 and Guba et al. While a small overestimation of the number of possible conformers in chains is accepted, this is not defensible when considering ring structures. A quick estimation of the overcount shows the importance of taking correlation into account for rings: Given the SMARTS pattern matching the dihedrals in an aliphatic six-membered ring, there would be three bins per bond, which leads to 3 6 = 729 possible combinations of the six bits of the TABS for the ring. For the purposes of the analysis in this study, we used nTABS to further decompose the Platinum set into three subsets: low flexibility molecules (nTABS < 500), medium flexibility molecules (500 < = nTABS < 10,000), and high flexibility molecules (nTABS > = 10,000). Note on Calculating TABS with ETKDG In order to generate a TABS for a As this library does not cover all bonds assigned to be rotatable using the RDKit’s definition, 15 we need to calculate multiplicities and bins also for these additional rotatable bonds. The additional dihedrals were assigned a multiplicity of six as a default option, binning the torsion profile arbitrarily at 30, 90, 150, 210, 270 and 330°. Analysis For the comparison of the categorization of conformers with TABS and with heavy-atom RMSD, we calculated confusion matrices.