Symptom scores vs fungal composition (Spearman and PERMANOVA)

Survey symptom scores (HBI and related items) tested against alpha diversity (Spearman) and genus beta diversity (PERMANOVA on Bray–Curtis).

1 Research question

Are symptom scores from participant surveys associated with fungal composition (alpha and beta diversity)?

Null (Spearman): No monotonic association between each symptom score and each alpha diversity metric.

Null (PERMANOVA): Genus-level community composition is not associated with the symptom score after accounting for Bray–Curtis distance structure.

2 Data

Role	Source
Mycobiome (alpha)	data/intermediate/alpha_long.rds (make wrangle)
Mycobiome (genus beta)	data/processed/genus.csv (make save)
Survey / symptoms	data/processed/cleaned_characteristics.csv (make characteristics on data/raw/SYN_Participant Characteristics(Sheet1).csv)

• Unit of analysis: Stool sample (16 samples, 12 participants with mycobiome data).
• Join key: Participant_ID (mycobiome) = participant_id (characteristics); one characteristics row per participant used when duplicates exist.
• Symptom scores: Numeric fields and ordinally coded items parsed to numbers (e.g. Mild = 1, 4\tSOME OF THE TIME → 4). Primary focus: Harvey–Bradshaw Index (HBI); additional exploratory symptoms in the correlation table.
• Inclusion: Samples retained after 02_data_wrangling.R removals; PERMANOVA genus analysis uses samples with non-missing HBI.

Coverage: 16 samples, 12 participants; 16 samples with non-missing HBI.

3 Methods

Item	Spearman (symptoms × alpha)	PERMANOVA (symptoms × taxa)
Test	Spearman rank correlation (cor.test, method = "spearman")	vegan::adonis2
Response	Shannon, Simpson, Chao1	Genus Bray–Curtis dissimilarity
Predictor	Parsed symptom scores	Harvey–Bradshaw Index (continuous)
Multiple testing	Benjamini–Hochberg across all symptom × metric pairs	Exploratory single predictor
Permutations	—	999
Significance	α = 0.05	α = 0.05

spearman_results <- spearman_symptom_alpha(sample_symptoms)

permanova_hbi <- run_permanova_symptom(
  genus_wide,
  prefix = TAXA_LEVELS$genus$prefix,
  symptom_col = "harvey_bradshaw_index"
)

4 Assumptions and diagnostics

• Independence: Samples are not fully independent (up to two stool replicates per participant). Results are exploratory; consider participant-level sensitivity analyses if reporting formally.
• Spearman: Tests monotonic association; does not assume linearity or normality.
• PERMANOVA: Exchangeability under permutations; composition constrained to genus relative abundance (Bray–Curtis).
• Missing symptoms: HBI and other scores are not available for all participants (see coverage above).

5 Results

5.1 Spearman: symptom scores × alpha diversity

Symptom	Alpha metric	n	Spearman rho	p	BH-adjusted p
Harvey-Bradshaw Index	Shannon	16	-0.4543	0.0771	0.6620
Harvey-Bradshaw Index	Simpson	16	0.2331	0.3849	0.7339
Harvey-Bradshaw Index	Chao1	16	0.3595	0.1715	0.6620
Daily soft stools	Shannon	16	0.2205	0.4118	0.7339
Daily soft stools	Simpson	16	-0.0214	0.9372	0.9406
Daily soft stools	Chao1	16	-0.2757	0.3014	0.6782
Abdominal pain	Shannon	16	-0.3009	0.2574	0.6620
Abdominal pain	Simpson	16	-0.0203	0.9406	0.9406
Abdominal pain	Chao1	16	-0.0483	0.8589	0.9406
Fatigue frequency	Shannon	16	0.3156	0.2337	0.6620
Fatigue frequency	Simpson	16	-0.1593	0.5556	0.7339
Fatigue frequency	Chao1	16	-0.1683	0.5332	0.7339
Anxiety frequency	Shannon	16	-0.0932	0.7314	0.8777
Anxiety frequency	Simpson	16	0.1533	0.5708	0.7339
Anxiety frequency	Chao1	16	-0.3367	0.2023	0.6620
Abdominal bloating frequency	Shannon	16	-0.1894	0.4824	0.7339
Abdominal bloating frequency	Simpson	16	-0.3277	0.2154	0.6620
Abdominal bloating frequency	Chao1	16	0.3848	0.1411	0.6620

No symptom × alpha pairs were significant after BH adjustment (q < 0.05).

5.2 PERMANOVA: genus composition ~ HBI

Term	Df	Sum of squares	R²	F	Pr(>F)
Model	1	0.1006	0.059	0.8786	0.658
Residual	14	1.6035	0.941	NA	NA
Total	15	1.7041	1.000	NA	NA

Harvey–Bradshaw Index explained 5.9% of variance in genus composition (R² = 0.059; pseudo-F = 0.88; p = 0.6580, 999 permutations; n = 16 samples with HBI).

6 Figures

6.1 HBI vs Shannon diversity (Spearman)

plot_symptom_alpha_scatter(
  sample_symptoms,
  symptom_col = "harvey_bradshaw_index",
  metric = "Shannon"
)

Figure 1: Harvey–Bradshaw Index vs Shannon diversity by study group. Subtitle shows Spearman correlation across samples with complete data.

6.2 Genus PCoA coloured by HBI (PERMANOVA)

plot_pcoa_symptom(
  permanova_hbi,
  TAXA_LEVELS$genus$label
)

Figure 2: PCoA of genus Bray–Curtis dissimilarity; points coloured by Harvey–Bradshaw Index.

7 Interpretation

Spearman correlations describe pairwise monotonic links between symptom scores and alpha metrics at the sample level; they do not adjust for repeated measures per participant. PERMANOVA tests whether multivariate genus composition covaries with HBI across samples; a non-significant p-value does not rule out associations with individual taxa or other symptom instruments.

Correlation among symptoms and small sample size limit power; treat findings as exploratory unless pre-registered.

8 Reproducibility

• Helpers: stats/R/symptom_association_helpers.R, stats/R/permanova_helpers.R
• Optional merged table: make merge builds data/processed/merged.csv for dashboard use; this post joins the same sources directly.
• Render: quarto render stats from repository root