Editing Quantitative structure–activity relationship (section)

== Types ==

=== Fragment based (group contribution) ===

Analogously, the "[[partition coefficient]]"—a measurement of differential solubility and itself a component of QSAR predictions—can be predicted either by atomic methods (known as "XLogP" or "ALogP") or by [[group contribution method|chemical fragment methods]] (known as "CLogP" and other variations). It has been shown that the [[partition coefficient|logP]] of compound can be determined by the sum of its fragments; fragment-based methods are generally accepted as better predictors than atomic-based methods.<ref name="pmid17597897">{{cite journal | vauthors = Thompson SJ, Hattotuwagama CK, Holliday JD, Flower DR | title = On the hydrophobicity of peptides: Comparing empirical predictions of peptide log P values | journal = Bioinformation | volume = 1 | issue = 7 | pages = 237–41 | year = 2006 | pmid = 17597897 | pmc = 1891704 | doi = 10.6026/97320630001237 }}</ref> Fragmentary values have been determined statistically, based on empirical data for known logP values. This method gives mixed results and is generally not trusted to have accuracy of more than ±0.1 units.<ref>{{Cite journal | title = Prediction of physicochemical parameters by atomic contributions |vauthors=Wildman SA, Crippen GM | doi = 10.1021/ci990307l | year = 1999 | journal = J. Chem. Inf. Comput. Sci. | pages = 868–873 | volume = 39 | issue = 5 }}</ref>

Group or fragment-based QSAR is also known as GQSAR.<ref name="Ajmani_2008"/> GQSAR allows flexibility to study various molecular fragments of interest in relation to the variation in biological response. The molecular fragments could be substituents at various substitution sites in congeneric set of molecules or could be on the basis of pre-defined chemical rules in case of non-congeneric sets.  GQSAR also considers cross-terms fragment descriptors, which could be helpful in identification of key fragment interactions in determining variation of activity.<ref name="Ajmani_2008">{{citation | vauthors = Ajmani S, Jadhav K, Kulkarni SA | title = Group-Based QSAR (G-QSAR)}}</ref>
Lead discovery using fragnomics is an emerging paradigm. In this context FB-QSAR proves to be a promising strategy for fragment library design and in fragment-to-lead identification endeavours.<ref>{{cite journal | vauthors = Manoharan P, Vijayan RS, Ghoshal N | title = Rationalizing fragment based drug discovery for BACE1: insights from FB-QSAR, FB-QSSR, multi objective (MO-QSPR) and MIF studies | journal = Journal of Computer-Aided Molecular Design | volume = 24 | issue = 10 | pages = 843–64 | date = Oct 2010 | pmid = 20740315 | doi = 10.1007/s10822-010-9378-9 | bibcode = 2010JCAMD..24..843M | s2cid = 1171860 }}</ref>

An advanced approach on fragment or group-based QSAR based on the concept of pharmacophore-similarity is developed.<ref name ="Kumar_2013"/> This method, pharmacophore-similarity-based QSAR (PS-QSAR) uses topological pharmacophoric descriptors to develop QSAR models. This activity prediction may assist the contribution of certain pharmacophore features encoded by respective fragments toward activity improvement and/or detrimental effects.<ref name="Kumar_2013">{{cite journal | vauthors = Prasanth Kumar S, Jasrai YT, Pandya HA, Rawal RM | title = Pharmacophore-similarity-based QSAR (PS-QSAR) for group-specific biological activity predictions | journal = Journal of Biomolecular Structure & Dynamics | volume = 33 | issue = 1 | pages = 56–69 | date = November 2013 | pmid = 24266725 | doi = 10.1080/07391102.2013.849618 | s2cid = 45364247 | url = https://figshare.com/articles/dataset/Pharmacophore_similarity_based_QSAR_PS_QSAR_for_group_specific_biological_activity_predictions/861021 | url-access = subscription }}</ref>

=== 3D-QSAR ===

The acronym '''3D-QSAR''' or '''3-D QSAR''' refers to the application of [[Force field (chemistry)|force field]] calculations requiring three-dimensional structures of a given set of small molecules with known activities (training set). The training set needs to be superimposed (aligned) by either experimental data (e.g. based on ligand-protein [[crystallography]]) or molecule [[superimposition]] software. It uses computed potentials, e.g. the [[Lennard-Jones potential]], rather than experimental constants and is concerned with the overall molecule rather than a single substituent. The first 3-D QSAR was named Comparative Molecular Field Analysis (CoMFA) by Cramer et al. It examined the steric fields (shape of the molecule) and the electrostatic fields<ref name="isbn0-582-38210-6">{{cite book | vauthors = Leach AR | title = Molecular modelling: principles and applications | publisher = Prentice Hall | location = Englewood Cliffs, N.J | year = 2001  | isbn = 978-0-582-38210-7 }}</ref> which were correlated by means of [[partial least squares regression]] (PLS).

The created data space is then usually reduced by a following [[feature extraction]] (see also [[dimensionality reduction]]). The following learning method can be any of the already mentioned [[machine learning]] methods, e.g. [[support vector machine]]s.<ref name="isbn0-262-19509-7">{{cite book | vauthors = Vert JP, Schölkopf B, Tsuda K | title = Kernel methods in computational biology | publisher = MIT Press | location = Cambridge, Mass | year = 2004  | isbn = 978-0-262-19509-6 }}</ref> An alternative approach uses [[multiple-instance learning]] by encoding molecules as sets of data instances, each of which represents a possible molecular conformation. A label or response is assigned to each set corresponding to the activity of the molecule, which is assumed to be determined by at least one instance in the set (i.e. some conformation of the molecule).<ref>{{cite journal | vauthors = Dietterich TG, Lathrop RH, Lozano-Pérez T | title = Solving the multiple instance problem with axis-parallel rectangles | journal = Artificial Intelligence | volume = 89 | issue = 1–2 | year = 1997 | pages = 31–71 | doi = 10.1016/S0004-3702(96)00034-3}}</ref>

On June 18, 2011 the Comparative Molecular Field Analysis (CoMFA) patent has dropped any restriction on the use of GRID and partial least-squares (PLS) technologies.{{citation needed|date=March 2018}}

=== Chemical descriptor based ===

In this approach, descriptors quantifying various electronic, geometric, or steric properties of a molecule are computed and used to develop a QSAR.<ref>{{cite journal | vauthors = Caruthers JM, Lauterbach JA, Thomson KT, Venkatasubramanian V, Snively CM, Bhan A, Katare S, Oskarsdottir G | title = Catalyst design: knowledge extraction from high-throughput experimentation | journal = J. Catal. | year = 2003 | volume = 216 | issue = 1–2 | pages = 3776–3777 | doi = 10.1016/S0021-9517(02)00036-2}}</ref> This approach is different from the fragment (or group contribution) approach in that the descriptors are computed for the system as whole rather than from the properties of individual fragments. This approach is different from the 3D-QSAR approach in that the descriptors are computed from scalar quantities (e.g., energies, geometric parameters) rather than from 3D fields.

An example of this approach is the QSARs developed for olefin polymerization by [[half sandwich compound]]s.<ref name="pmid17348648">{{cite journal | vauthors = Manz TA, Phomphrai K, Medvedev G, Krishnamurthy BB, Sharma S, Haq J, Novstrup KA, Thomson KT, Delgass WN, Caruthers JM, Abu-Omar MM | title = Structure-activity correlation in titanium single-site olefin polymerization catalysts containing mixed cyclopentadienyl/aryloxide ligation | journal = Journal of the American Chemical Society | volume = 129 | issue = 13 | pages = 3776–7 | date = Apr 2007 | pmid = 17348648 | doi = 10.1021/ja0640849 }}</ref><ref name = "Organometallics2012">{{cite journal | vauthors = Manz TA, Caruthers JM, Sharma S, Phomphrai K, Thomson KT, Delgass WN, Abu-Omar MM | title = Structure–Activity Correlation for Relative Chain Initiation to Propagation Rates in Single-Site Olefin Polymerization Catalysis | journal = Organometallics | year = 2012 | volume = 31 | pages = 602–618 | doi = 10.1021/om200884x | issue = 2}}</ref>

=== String based ===
It has been shown that activity prediction is even possible based purely on the [[Simplified Molecular Input Line Entry Specification|SMILES]] string.<ref>{{cite arXiv |last1=Jastrzębski |first1=Stanisław |last2=Leśniak |first2=Damian |last3=Czarnecki |first3=Wojciech Marian |title=Learning to SMILE(S)  |date=8 March 2018 |class=cs.CL |eprint=1602.06289}}</ref><ref>{{cite arXiv |last1=Bjerrum |first1=Esben Jannik |title=SMILES Enumeration as Data Augmentation for Neural Network Modeling of Molecules  |date=17 May 2017 |class=cs.LG |eprint=1703.07076}}</ref><ref>{{cite journal |last1=Mayr |first1=Andreas |last2=Klambauer |first2=Günter |last3=Unterthiner |first3=Thomas |last4=Steijaert |first4=Marvin |last5=Wegner |first5=Jörg K. |last6=Ceulemans |first6=Hugo |last7=Clevert |first7=Djork-Arné |last8=Hochreiter |first8=Sepp |title=Large-scale comparison of machine learning methods for drug target prediction on ChEMBL |journal=Chemical Science |date=20 June 2018 |volume=9 |issue=24 |pages=5441–5451 |doi=10.1039/c8sc00148k|pmid=30155234 |pmc=6011237 }}</ref>

=== Graph based ===
Similarly to string-based methods, the molecular graph can directly be used as input for QSAR models,<ref>{{cite journal |last1=Merkwirth |first1=Christian |last2=Lengauer |first2=Thomas |title=Automatic Generation of Complementary Descriptors with Molecular Graph Networks |journal=Journal of Chemical Information and Modeling |date=1 September 2005 |volume=45 |issue=5 |pages=1159–1168 |doi=10.1021/ci049613b|pmid=16180893 }}</ref><ref>{{cite journal |last1=Kearnes |first1=Steven |last2=McCloskey |first2=Kevin |last3=Berndl |first3=Marc |last4=Pande |first4=Vijay |last5=Riley |first5=Patrick |title=Molecular graph convolutions: moving beyond fingerprints |journal=Journal of Computer-Aided Molecular Design |date=1 August 2016 |volume=30 |issue=8 |pages=595–608 |doi=10.1007/s10822-016-9938-8|pmid=27558503 |pmc=5028207 |arxiv=1603.00856 |bibcode=2016JCAMD..30..595K }}</ref> but usually yield inferior performance compared to descriptor-based QSAR models.<ref>{{cite journal |last1=Jiang |first1=Dejun |last2=Wu |first2=Zhenxing |last3=Hsieh |first3=Chang-Yu |last4=Chen |first4=Guangyong |last5=Liao |first5=Ben |last6=Wang |first6=Zhe |last7=Shen |first7=Chao |last8=Cao |first8=Dongsheng |last9=Wu |first9=Jian |last10=Hou |first10=Tingjun |title=Could graph neural networks learn better molecular representation for drug discovery? A comparison study of descriptor-based and graph-based models |journal=Journal of Cheminformatics |date=17 February 2021 |volume=13 |issue=1 |pages=12 |doi=10.1186/s13321-020-00479-8|pmid=33597034 |pmc=7888189 |doi-access=free }}</ref><ref>{{cite journal |last1=van Tilborg |first1=Derek |last2=Alenicheva |first2=Alisa |last3=Grisoni |first3=Francesca |title=Exposing the Limitations of Molecular Machine Learning with Activity Cliffs |journal=Journal of Chemical Information and Modeling |date=12 December 2022 |volume=62 |issue=23 |pages=5938–5951 |doi=10.1021/acs.jcim.2c01073|pmid=36456532 |pmc=9749029 }}</ref>

===q-RASAR===
QSAR has been merged with the similarity-based read-across technique to develop a new field of [[q-RASAR]]. The [https://sites.google.com/site/kunalroyindia/home/the-dtc-laboratory?authuser=0 DTC Laboratory] at [[Jadavpur University]] has developed this hybrid method and the details are available at their [https://sites.google.com/site/kunalroyindia/home/rasar?authuser=0 laboratory page]. Recently, the q-RASAR framework has been improved by its integration with the [[ARKA descriptors in QSAR]].