Generative modeling for protein structures

Authors: Namrata Anand, Possu Huang

NeurIPS 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We test the effectiveness of our models by predicting completions of corrupted protein structures and show that the method is capable of quickly producing structurally plausible solutions. We use data from the Protein Data Bank [30], a repository of experimentally determined structures available on-line. We generated 16-, 64-, and 128-residue maps by training GANs on the corresponding maps in our dataset. We compare our method for structure generation to the following baselines: Hidden Markov Model (HMM) based methods Torus DBN [17] and FB5-HMM [18], a multi-scale torsion angle GAN, 3DGAN [29], and a full-atom GAN (2D pairwise distance maps for full-atom peptide backbones).
Researcher Affiliation Academia Namrata Anand Bioengineering Department, Stanford namrataa@stanford.edu Po-Ssu Huang Bioengineering Department, Stanford possu@stanford.edu
Pseudocode No The paper presents mathematical formulations of algorithms (e.g., ADMM updates) but does not include structured pseudocode or an algorithm block.
Open Source Code No The paper does not contain an explicit statement about releasing source code or a direct link to a code repository for the described methodology.
Open Datasets Yes We use data from the Protein Data Bank [30], a repository of experimentally determined structures available on-line.
Dataset Splits No The paper mentions separating 'train and test structures' but does not specify a validation dataset split or provide details for creating one.
Hardware Specification No The paper does not provide specific details about the hardware (e.g., GPU/CPU models, memory specifications) used to run the experiments.
Software Dependencies Yes SCS: Splitting conic solver, version 2.0.2. https://github.com/cvxgrp/scs, November 2017.
Experiment Setup No The paper states, 'Experiment details are given in the supplementary material, Section A.1,' but does not provide specific hyperparameters or detailed training configurations within the main text.