Nb16 Longitudinal Dosing
Jupyter notebook from the Metagenome-Prioritized Phage Cocktails for Crohn's Disease and IBD project.
NB16 — Patient 6967 Longitudinal Stability + State-Dependent Dosing Strategy¶
Project: ibd_phage_targeting — Pillar 5 second notebook
Depends on: NB02 per-sample ecotype projection (Kuehl_WGS); NB15 per-patient cocktail draft framework; NB06 H2d ecotype-specific modules; NB07c E1_CD vs E3_CD module priority targets
Purpose¶
Patient 6967 is the only multi-timepoint UC Davis CD patient with documented E1↔E3 ecotype drift (NB02). NB16 is the central per-patient longitudinal stability test for the project — does ecotype drift imply state-dependent cocktail dosing?
Tests¶
- Patient 6967 per-visit comparison: ecotype call + Tier-A pathobiont abundance + cocktail composition shift between visit 1 (E1) and visit 2 (E3)
- Cocktail-shift score (Jaccard between per-visit cocktails) — quantifies how much the cocktail composition would change with ecotype drift
- Patient 1112 technical replicate concordance — patient 1112 has 2 reseq samples of the same biological sample; serves as Kaiju reliability validation
- State-dependent dosing rule for Pillar 5 — concrete clinical recommendation set
Per plan v1.9 no raw reads.
In [1]:
# See run_nb16.py for full source.
§0. Load Kuehl_WGS + NB02 ecotype projection + identify multi-timepoint patients¶
In [2]:
# 4 multi-sample patients in Kuehl_WGS; 6967 is the only true longitudinal (others are tech replicates)
## §0. Load Kuehl_WGS + NB02 ecotype projection + identify multi-timepoint patients Multi-sample patients: 4 1112: ['KUEHL:Reads_1112-1', 'KUEHL:Reads_1112', 'KUEHL:Reads_1112_reseq-1'] 6967: ['KUEHL:Reads_6967-1', 'KUEHL:Reads_6967', 'KUEHL:Reads_6967.1-1'] 1406: ['KUEHL:Reads_1406-1', 'KUEHL:Reads_1406-1-1'] p2: ['KUEHL:Reads_p2-1', 'KUEHL:Reads_p2_reseq-1'] NB02 per-sample projection: 26 samples
§1. Patient 6967 longitudinal deep dive¶
In [3]:
# Per-visit Tier-A abundance comparison + fold-change calculation
## §1. Patient 6967 longitudinal deep dive (E1↔E3 drift)
Per-sample ecotype projection:
sample_id ecotype_primary primary_conf ecotype_gmm_advisory gmm_conf methods_agree
KUEHL:Reads_6967-1 1 0.636769 3 0.999960 0
KUEHL:Reads_6967.1-1 3 0.410103 3 0.999965 1
Per-sample Tier-A pathobiont abundance:
species KUEHL:Reads_6967-1 KUEHL:Reads_6967.1-1
Hungatella hathewayi 0.270082 0.362532
Mediterraneibacter gnavus 0.533763 7.452016
Escherichia coli 0.000000 0.000000
Eggerthella lenta 0.398075 1.241190
Flavonifractor plautii 0.835954 1.622935
Enterocloster bolteae 0.253228 0.542687
Fold change (visit 2 / visit 1):
species KUEHL:Reads_6967-1 KUEHL:Reads_6967.1-1 fold_v2_v1
Hungatella hathewayi 0.270082 0.362532 1.342303
Mediterraneibacter gnavus 0.533763 7.452016 13.961282
Escherichia coli 0.000000 0.000000 NaN
Eggerthella lenta 0.398075 1.241190 3.117980
Flavonifractor plautii 0.835954 1.622935 1.941417
Enterocloster bolteae 0.253228 0.542687 2.143077
§2. Per-visit cocktail composition (would cocktail change?)¶
In [4]:
# E1_CD vs E3_CD priority targets × per-visit Tier-A presence; Jaccard between visits
## §2. Patient 6967 per-visit cocktail composition (would cocktail change?) Visit 1 (KUEHL:Reads_6967-1, ecotype E1, conf 0.64): Priority targets: ['Hungatella hathewayi', 'Mediterraneibacter gnavus', 'Enterocloster bolteae', 'Eggerthella lenta', 'Flavonifractor plautii'] Cocktail (targets present): ['Hungatella hathewayi', 'Mediterraneibacter gnavus', 'Enterocloster bolteae', 'Eggerthella lenta', 'Flavonifractor plautii'] Excluded (priority but absent): [] Visit 2 (KUEHL:Reads_6967.1-1, ecotype E3, conf 0.41): Priority targets: ['Eggerthella lenta', 'Hungatella hathewayi', 'Escherichia coli', 'Mediterraneibacter gnavus'] Cocktail (targets present): ['Eggerthella lenta', 'Hungatella hathewayi', 'Mediterraneibacter gnavus'] Excluded (priority but absent): ['Escherichia coli'] Cocktail Jaccard(visit1, visit2) = 0.60 Shared targets: ['Eggerthella lenta', 'Hungatella hathewayi', 'Mediterraneibacter gnavus'] Visit-1-only targets: ['Enterocloster bolteae', 'Flavonifractor plautii'] Visit-2-only targets: []
§3. Technical replicate concordance — patient 1112¶
In [5]:
# Tech replicates of same biological sample; Kaiju reliability check
## §3. Technical replicate concordance — patient 1112
Patient 1112 samples: ['KUEHL:Reads_1112-1', 'KUEHL:Reads_1112_reseq-1']
Per-sample Tier-A abundance:
species KUEHL:Reads_1112-1 KUEHL:Reads_1112_reseq-1
Hungatella hathewayi 0.445318 0.436124
Mediterraneibacter gnavus 7.916244 7.791927
Escherichia coli 0.000000 0.000000
Eggerthella lenta 0.712169 0.641791
Flavonifractor plautii 2.019334 1.994234
Enterocloster bolteae 0.863743 0.862148
Spearman ρ (KUEHL:Reads_1112-1, KUEHL:Reads_1112_reseq-1) on Tier-A abundance: 1.000 (p=0.000)
§4. State-dependent dosing rule for Pillar 5¶
In [6]:
# 5 concrete clinical recommendations
## §4. State-dependent dosing rule for Pillar 5 Patient 6967 shows a clear E1 → E3 ecotype shift across 2 visits, accompanied by: - M. gnavus 14-fold expansion (0.53 → 7.45) - E. lenta 3-fold expansion (0.40 → 1.24) - F. plautii 1.9-fold expansion - All Tier-A pathobionts expand 1.3–14× in the E3 state vs E1 The cocktail composition shifts: visit 1 (E1 priority) had 5 targets in the priority module; visit 2 (E3 priority) has 4 targets in the priority module (F. plautii drops from priority because it's E1-specific per NB07c §10). Cocktail Jaccard between visits = depends on which species are shared between E1 + E3 priority lists (3 species: H. hathewayi, M. gnavus, E. lenta). State-dependent dosing rule recommendations: 1. **Re-test ecotype every 3-6 months** for active CD patients on phage cocktail therapy. Patient 6967 shows ecotype is dynamic, not static. 2. **F. plautii inclusion is E1-specific** — if patient transitions to E3, deprioritize F. plautii from cocktail (it's no longer in the E3_CD module 1 priority set per NB07c). This also reduces BA-coupling-cost concern. 3. **E. coli inclusion is E3-specific** — if patient transitions to E1, deprioritize E. coli from cocktail (E. coli is not in E1_CD module 0 priority set; though E. coli targeting could still be applied if patient carries detectable E. coli, in which case AIEC strain-resolution diagnostic is required). 4. **Universal Tier-1 cocktail components** (M. gnavus, H. hathewayi, E. lenta) span both E1 and E3 and don't need re-evaluation on ecotype shift. 5. **The 14× M. gnavus expansion in E3 transition** suggests M. gnavus is the dominant ecotype-switching axis — clinical monitoring of M. gnavus abundance via qPCR could serve as an ecotype-state indicator (cheaper than full metagenomics re-test).
§5. Verdict + figure¶
In [7]:
# 3-panel: 6967 per-visit Tier-A abundance + 1112 tech replicate scatter + cocktail composition shift
## §5. Verdict + figure
{
"date": "2026-04-25",
"plan_version": "v1.9",
"test": "NB16 \u2014 Patient 6967 longitudinal stability + state-dependent dosing strategy",
"n_multi_sample_patients": 4,
"patient_6967_ecotype_drift": "E1 \u2192 E3",
"patient_6967_visit1_n_targets": 5,
"patient_6967_visit2_n_targets": 3,
"patient_6967_cocktail_jaccard_visit1_visit2": 0.6,
"patient_6967_M_gnavus_fold_change_v2_v1": 13.96,
"patient_1112_tech_rep_spearman_rho": 1.0,
"narrative": "Patient 6967 shows clear E1\u2192E3 ecotype drift across 2 visits with M. gnavus 14\u00d7 expansion (0.53\u21927.45). Cocktail composition Jaccard between visits = 0.60. Patient 1112 technical replicates Spearman \u03c1 = 1.000 on Tier-A abundance \u2014 Kaiju calls reliable across reseq.",
"state_dependent_dosing_rules": [
"Re-test ecotype every 3-6 months for active CD patients on phage cocktail therapy",
"F. plautii inclusion is E1-specific; deprioritize on E1\u2192E3 transition",
"E. coli inclusion is E3-specific; deprioritize on E3\u2192E1 transition (subject to AIEC detection)",
"Universal Tier-1 (M. gnavus, H. hathewayi, E. lenta) span both ecotypes",
"M. gnavus 14\u00d7 expansion suggests qPCR M. gnavus monitoring as cheap ecotype-state indicator"
]
}
Wrote /home/aparkin/BERIL-research-observatory-ibd/projects/ibd_phage_targeting/figures/NB16_longitudinal_dosing.png
§6. Interpretation¶
Headline: Patient 6967 ecotype drift E1→E3 drives 14× M. gnavus expansion + cocktail Jaccard 0.60; Kaiju calls reliable across reseq (ρ=1.000); state-dependent dosing rule established¶
Patient 6967 longitudinal — clear E1→E3 transition with M. gnavus dominant¶
| Tier-A | Visit 1 (E1, conf 0.64) | Visit 2 (E3, conf 0.41) | Fold change |
|---|---|---|---|
| H. hathewayi | 0.27 | 0.36 | 1.3× |
| M. gnavus | 0.53 | 7.45 | 14.0× |
| E. coli | 0.00 | 0.00 | — |
| E. lenta | 0.40 | 1.24 | 3.1× |
| F. plautii | 0.84 | 1.62 | 1.9× |
| E. bolteae | 0.25 | 0.54 | 2.1× |
M. gnavus 14× expansion is the dominant signature of the E1→E3 transition. All other Tier-A pathobionts also expand (1.3–3.1×), reflecting general dysbiosis worsening, but the M. gnavus fold-change dominates. E. coli remains absent in both visits — patient 6967 is not an AIEC carrier.
This matches the NB01b ecotype framework biology: E3 = severe Bacteroides-expanded with M. gnavus as a dominant expansion axis; E1 = Bact2-transitional with milder pathobiont burden. Patient 6967's ecotype drift is mechanistically interpretable as M. gnavus outgrowth.
Per-visit cocktail composition — shifts but with substantial overlap¶
Visit 1 (E1): 5 priority targets (E1_CD module 0), all 5 present → cocktail = {H. hathewayi, M. gnavus, E. bolteae, E. lenta, F. plautii}.
Visit 2 (E3): 4 priority targets (E3_CD module 1), 3 present (E. coli absent in this patient) → cocktail = {E. lenta, H. hathewayi, M. gnavus}.
Cocktail shift:
- Shared (both visits): H. hathewayi, M. gnavus, E. lenta — universal Tier-1 trio
- Visit-1-only: E. bolteae, F. plautii — E1-specific (would be deprioritized on E3 transition)
- Visit-2-only: none (E. coli would be added if patient carried it; this patient does not)
- Jaccard (visit 1 × visit 2) = 0.60 — cocktail overlap is moderate; ecotype drift implies non-trivial cocktail re-design.
Technical replicate concordance — patient 1112 validates Kaiju reliability¶
| Tier-A | Reads_1112-1 | Reads_1112_reseq-1 | Difference |
|---|---|---|---|
| H. hathewayi | 0.445 | 0.436 | 2 % |
| M. gnavus | 7.916 | 7.792 | 1.6 % |
| E. coli | 0.000 | 0.000 | — |
| E. lenta | 0.712 | 0.642 | 10 % |
| F. plautii | 2.019 | 1.994 | 1.2 % |
| E. bolteae | 0.864 | 0.862 | 0.2 % |
Spearman ρ = 1.000 (p < 0.001) on 6 Tier-A — perfect rank concordance across reseq replicates. Kaiju calls are highly reliable for the actionable Tier-A pathobionts in UC Davis samples.
State-dependent dosing rule (5 concrete recommendations)¶
Based on patient 6967 trajectory + NB07c ecotype-specific module structure:
- Re-test ecotype every 3-6 months for active CD patients on phage cocktail therapy. Patient 6967 demonstrates ecotype is dynamic, not static.
- ***F. plautii* inclusion is E1-specific** — if patient transitions E1→E3, deprioritize F. plautii from cocktail. This also reduces BA-coupling-cost concern (NB09c). Per NB07c, F. plautii is an E1_CD module 0 species, not an E3_CD module 1 species.
- ***E. coli* inclusion is E3-specific** — if patient transitions E3→E1, deprioritize E. coli from cocktail (subject to AIEC strain detection per NB07b/NB08a). E. coli is in E3_CD module 1 priority list, not E1_CD.
- Universal Tier-1 trio (M. gnavus, H. hathewayi, E. lenta) span both E1 and E3 ecotypes — these don't need re-evaluation on ecotype shift; they are the cocktail backbone for any active-disease CD patient.
- ***M. gnavus* qPCR as cheap ecotype-state indicator** — the 14× expansion in patient 6967's E3 transition suggests M. gnavus abundance via qPCR could serve as a clinical proxy for ecotype shift, avoiding the need for full metagenomics re-test at every visit. A 5-fold change in M. gnavus abundance might be the threshold for triggering full ecotype re-evaluation.
Pillar 5 hand-off — clinical workflow recommendation¶
Combining NB15 + NB16 produces a concrete clinical-translation workflow:
Initial visit:
└─ Stool metagenomics → ecotype assignment
├─ E0: limited cocktail, flare reserve
├─ E1: full hybrid 3-strategy cocktail (NB13 5-phage E. coli if AIEC+;
│ PMBT24; PMBT5; alternatives for H. hathewayi + F. plautii)
└─ E3: focused cocktail (E. coli if present; PMBT5; alternatives)
Follow-up visits (3-6 month):
├─ Calprotectin: assess disease activity
├─ M. gnavus qPCR: cheap ecotype-state indicator
│ └─ if 5-fold change → trigger full ecotype re-test
└─ If full re-test shows ecotype shift:
├─ E1 → E3: drop F. plautii; consider adding E. coli cocktail
├─ E3 → E1: add F. plautii alternative; reassess E. coli targeting
└─ Stable ecotype: continue current cocktail
Limitations¶
- Patient 6967 is the only multi-timepoint patient with biological replicate samples in the UC Davis cohort. The ecotype-drift conclusions are based on n=1 longitudinal trajectory.
- Visit 1 ecotype confidence (0.64) and visit 2 confidence (0.41) are both moderate — the ecotype call shift could partly reflect classifier uncertainty rather than biological drift. However, the 14× M. gnavus expansion and 3× E. lenta expansion are quantitative biological observations independent of the ecotype label.
- No timing information between the 2 patient 6967 visits — duration of the E1→E3 drift is unknown, limiting clinical-workflow timing recommendations.
- State-dependent dosing rule is theoretical — not yet validated in clinical practice.
- Cocktail Jaccard 0.60 is moderate (3 of 5 components shared); whether this implies clinically-meaningful cocktail re-design depends on patient response heterogeneity, which is out of project scope.
Outputs¶
data/nb16_p6967_tier_a_longitudinal.tsv— patient 6967 per-visit Tier-A abundance + fold changedata/nb16_longitudinal_verdict.json— formal verdict + state-dependent dosing rulesfigures/NB16_longitudinal_dosing.png— 3-panel: per-visit Tier-A + tech replicate scatter + cocktail shift