Maximum Unbiased Validation (MUV) Data Sets for Virtual Screening Based on PubChem Bioactivity Data

By Sebastian G. Rohrer, Knut Baumann. 2009

375 citations recorded by the source corpus.

Subfields: Computational Theory and Mathematics, Molecular Biology, Evaluation & Benchmarks, Computational Chemistry & Drug Discovery

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Theory and Mathematics

In Molecular Biology

In Evaluation & Benchmarks

In Computational Chemistry & Drug Discovery

Citations in the corpus

Interpreting linear support vector machine models with heat map molecule coloring (2011, methodology): “First, we employed the 17 MUV data sets compiled by Rohrer et al. [11] with their corresponding background data sets.”
Learning Graph-Level Representation for Drug Discovery (2017, background): “MUV The MUV dataset(Rohrer and Baumann 2009) contains 17 challenging tasks for around 90,000 compounds and is specifically designed for virtual screening techniques.”
Massively Multitask Networks for Drug Discovery (2015, background): “For instance, care must be taken to guard against unrealistic divisions between active and inactive compounds (“artiﬁcial enrichment”) and against information leakage due to strong similarity between active compounds (“analog bias”) (Rohrer & Baumann, 2009).”
AtomNet: A Deep Convolutional Neural Network for Bioactivity Prediction in Structure-based Drug Discovery (2015, background): “The dependence on known active ligands also makes it difﬁcult to show that the network is “right for the right reasons”; artifacts in the training data, such as analogue bias, make it very difﬁcult to properly assess accuracy and generalizability [16, 17, 18].”
Directory of Useful Decoys, Enhanced (DUD-E): Better Ligands and Decoys for Better Benchmarking (2012, background): “36 Instead of generating decoys at all, REPROVIS-DB assembles ligand and database data from earlier successful virtual screens which are deemed reproducible.”
Pre-training Molecular Graph Representation with 3D Geometry (2021, neutral): “Molecular Property: Biophysics Maximum Unbiased Validation (MUV) [76] is another subdatabase from PCBA, and is obtained by applying a refined nearest neighbor analysis.”
Key Topics in Molecular Docking for Drug Design (2019, methodology): “The Maximum Unbiased Validation (MUV) [94] datasets were curated with special care for the chemical diversity of the actives set, in order to avoid over-representation of chemical entities and thus avert overestimation of performance.”
Relational Pooling for Graph Representations (2019, methodology): “Therefore, training the models on MUV was expensive and we were only able to run for 10 random data splits as opposed to 20 for the others.”
BERT Learns (and Teaches) Chemistry (2020, methodology): “Finally, we would like to transfer this knowledge to tasks on smaller, more difﬁcult datasets, such as the Tox21 [14] and MUV [26] multitask datasets.”
Benchmarking platform for ligand-based virtual screening (2016, methodology): “Our compiled benchmark consists of 7 existing sets of datasets: ULS and UDS Data Sets [16], GDD and GLL [17], MUBD-HDACs Data Sets [IS], DUD Release 2 [19], NRLiSt BDB [20], MUV [21], and dataset from Comparative analysis of pharmacophore”
Open-source platform to benchmark fingerprints for ligand-based virtual screening (2013, methodology): “Two recently developed and publicly available compound data sets do not have this problem: the directory of useful decoys (DUD) [22,23] and the maximum unbiased validation (MUV) data sets [24,25].”