A Computing Chemical Reaction Network based on Glyconitrile

Here we describe all the files and datasets included in this repository. Once the main aspects of glyconitrile reactivity had been initially set up in the initial exploratory phase using NMR, HPLC-MS and UV-Vis spectroscopy, the in-depth chemical analysis of the CRN was performed through a well-defined metabolomics protocol (see Methods and SI) applied to HPLC-MS datasets obtained from systematically explored matrices of conditions (described in Figs. S5 and S6). For each of these matrices, the repository presents a zip file with the raw HPLC-MS data and a csv file with the metadata (experiment name, temperature, NaOH conc., glyconitrile conc., NaCl conc., and reaction time). The explored matrices of conditions are named as: • Batch-Kinetic: corresponding to Fig. S5a.• Batch-3x3x3: corresponding to Fig. S5b. This dataset was analyzed in depth for the main text section describing the CRN non-linear, complex dynamics (Fig. 2 in the main text). • Batch-5x5: corresponding to Fig. S5c.• CSTR-3x5: corresponding to Fig. S6a.• CSTR-4x5: corresponding to Fig. S6b.• Batch-10x10: This dataset was specifically generated to demonstrate the non-linear behavior of the CRN (compound intensity heat maps in Fig. 2b in the main text) and for reservoir computing (Fig. 3c (ii)) in the main text upon varying temperature and NaOH concentration. The file “FT-IR_raw_data_40_238.csv” contains FT-IR raw spectra as acquired for 40 different input conditions as described in the main text. The data was used for solving the nonlinear classification tasks in Fig. 3c).The file “FT-IR_feature_data_40_238.csv” contains the features extracted from above FT-IR raw data as acquired for 40 different input conditions as described in the main text. The data was used for solving the nonlinear classification tasks in Fig. 3c).The file “SCINE-Exploration-Glyconitrile-CRN.gz” is a gz-compressed file containing a comprehensive list of database identifiers for all discovered compounds and key intermediates, together with the full database, and all relevant scripts for exploration and post-processing. The database can be loaded into the SCINE Heron graphical user interface to interactively navigate through the CRN.