Data for: PathMolD-AB: Spatiotemporal Pathways of Protein Folding using Parallel Molecular Dynamics with a Coarse-grained Model

Using the PathMolD-AB software, four datasets of protein folding trajectories were produced as case studies. Four proteins were simulated, one artificially created and three real-world proteins with a growing number of amino acids, the detailed as follows:

13FIBO: it has 13 amino acids, and was artificially created by \cite{stillinger1995collective}, by distributing the hydrophobic amino acids according to the Fibonacci sequence.
2GB1 (http://10.2210/pdb2GB1/pdb): this protein is in the group of the G proteins, which exerts signal transduction functions. The dysfunction of this protein is linked to diseases such as schizophrenia in humans;
1PLC (http://10.2210/pdb1PLC/pdb): this protein performs the function of electron transportation, which is related to the process of energy production in the cell. Its functional impairment results in cell death;
5NAZ (https://www.rcsb.org/structure/5NAZ): this is a globular structural protein of collagen, and it is related to the Goodpasture's and Alport's syndromes.

Link for the Dataset of Spatiotemporal Pathways of Protein Folding:
https://mega.nz/#F!C5QkHQ6A!Ng2xowc2hVPoHHiSB7ww-w

For each protein, a dataset was generated with 1,000 (for 13FIBO, 2GB1 and 1PLC) or 500 (for 5NAZ) different pathways. Due to the length of the last protein, fewer simulations were done. As mentioned before, all simulations start with structures randomly initialized in the 3D space, to achieve a high diversity of pathways, each one leading to the native conformation of the protein.

To guarantee a reliable stabilization of the native structure, the maximum number of time-steps (t_{max}) for the simulations of the 13FIBO, 2GB1 and 1PLC proteins were set to 3x 10^6 iterations and 1x 10^8 for the 5NAZ protein. Consequently, for standardizing the number of spatiotemporal states per pathway in each dataset, the step_size for 13FIBO, 2GB1 and 1PLC were 3000, and 8000 for the 5NAZ.
For each pathway, 1,000 folding states were recorded.

借助PathMolD-AB软件，本研究以蛋白质折叠轨迹为案例，成功构建了四个蛋白质折叠轨迹数据集。模拟的蛋白质包括一种人工合成的蛋白质及三种真实世界的蛋白质，其氨基酸数量依次增加，具体如下：

13FIBO：该蛋白质由13个氨基酸组成，系通过cite{stillinger1995collective}人工合成，其疏水性氨基酸的分布遵循斐波那契数列。
2GB1（http://10.2210/pdb2GB1/pdb）：此蛋白质属于G蛋白家族，负责信号转导功能。该蛋白质的功能障碍与人类的疾病，如精神分裂症相关；
1PLC（http://10.2210/pdb1PLC/pdb）：此蛋白质负责电子传递功能，与细胞能量产生过程相关。其功能受损可导致细胞死亡；
5NAZ（https://www.rcsb.org/structure/5NAZ）：该蛋白质为一种球状结构的胶原蛋白蛋白，与Goodpasture综合征和Alport综合征相关。

蛋白质折叠时空路径数据集的链接：https://mega.nz/#F!C5QkHQ6A!Ng2xowc2hVPoHHiSB7ww-w

对于每种蛋白质，均生成了包含1,000条（针对13FIBO、2GB1和1PLC）或500条（针对5NAZ）不同路径的数据集。由于最后一种蛋白质的长度较长，因此模拟的次数较少。如前所述，所有模拟均从随机初始化的3D空间结构开始，以实现路径的高多样性，每一条路径最终都导向蛋白质的天然构象。

为确保蛋白质天然结构的可靠稳定，13FIBO、2GB1和1PLC蛋白质的模拟最大时间步数（t_{max}）分别设置为3x 10^6次迭代和5NAZ的1x 10^8次迭代。因此，为标准化每个数据集中每条路径的空间时间状态数量，13FIBO、2GB1和1PLC的步长分别为3000，而5NAZ的步长为8000。

对于每条路径，记录了1,000个折叠状态。