*2.5. Processing and Analysis of 1H NMR Data*

All NMR spectra were phased, manually baseline corrected and exponentially linebroadened with a 1Hz window function using TOPSPIN 3.6.1 (Bruker) software. The spectra were then imported into Chenomx 8.4. A shim correction of 1.2 Hz was applied and the region from 4.56 ppm to 4.97 ppm was deleted to eliminate water resonance peaks. Peak assignment was performed using the Chenomx profiler tool fitting the spectral line to the proposed compounds in the standard Chenomx library. The efficacies of the extraction solvents, lysis methods and incubation temperatures were then compared using molecule concentration ratios and number of metabolites ratios between the two samples (Figure S3) (e.g.: IE/M = I ethanol/I methanol).

The median, standard deviation (StDev) and coefficient of variance (CV) were all calculated to determine the reproducibility of the results. Any outliers were detected and removed from the analysis.

#### **3. Results**

In order to determine the optimal protocol to extract metabolites from *Blastocystis* ST7 for NMR analysis, a series of extraction solvents, lysis techniques and incubation temperatures were examined. The efficacy of each protocol was assessed using proton NMR and the peak intensity was compared using TOPSPIN 3.6.1 to determine which method extracted the highest concentrations of metabolites. We then developed an efficient, reproducible protocol to perform metabolomics studies on *Blastocystis* species and found that the extraction solvent and lysis method were the most important factors for metabolite extraction. The efficacy was optimised in four sets of experiments, which firstly compared solvents (MeOH versus EtOH), then compared methods (sonication vs. bead bashing) and finally the temperature regime used (−20 ◦C versus RT,) and (60 ◦C versus RT).

#### *3.1. Comparison of Steps*

Two analysis methods were used during the comparisons of pairs of processing steps to rank efficacy. These were molar concentration ratios C μMA/B for processes A and B, as measured using the standard Chenomx metabolite library against the internal DSS standard. Secondly, the ratios of the raw number of detectable metabolites extracted NA/B using the two processes (A and B), again using the Chenomx metabolite library. All analyses were made in pairs of samples in triplicates, and samples in the triplicate were denoted by Arabic numerals 1–3 and condition pairs by A and B. Therefore, for a comparison of methanol and ethanol, 1A–3A were ethanol samples and 1B–3B were methanol samples.
