Two projects describe the same experimental phage-susceptibility matrix — a 96-phage × 188-strain × 17,672-pair inventory of binary infection outcomes (one susceptible/resistant call per strain–phage pair) 1,2. In the defense-system project the stated multiplication is internally inconsistent 2, while the phage-targeting project uses a 94-phage denominator (the count placed under the division bar when a proportion of phages is reported) elsewhere 1,2. Because the inconsistency is shared across reports rather than local to one, it cannot be read as a single report’s slip, and the inventory line cannot be taken as read wherever either report reuses it 1,2. The tension is recorded under adversarial-research-quality-assurance; no product is computed here, and the reported figures are retained exactly as stated.
Evidence Sides
The 96-phage inventory as stated (defense-system project). The defense-system project reports 188 E. coli strains (ECOR collection, NILS collection, clinical/environmental isolates) × 96 phages (41 Podoviridae, 32 Myoviridae, 23 Siphoviridae) = 17,672 binary infection outcomes, and that stated multiplication is internally inconsistent; the figures are retained as stated and no product is computed here. 2
The shared inventory with a 94-phage denominator (phage-targeting project). The same 96-phage × 188-strain × 17,672-pair inventory appears in the phage-targeting project, where a 94-phage denominator is used elsewhere, so the inconsistency is shared across reports rather than local to one. 1,2
Possible Reconciliations
- Hypothesis: 17,672 counts assayed pairs, not the full crossing. If some strain–phage combinations were never tested or failed quality control, the pair count would be a tested-pairs tally and the ”=” a reporting shorthand rather than arithmetic.
- Hypothesis: a filtering step separates 96 from 94. The two phage denominators may refer to different stages — phages deposited versus phages surviving a downstream filter — in which case each is correct for its own scope.
- Hypothesis: a single transcription error propagated by reuse. If one report’s inventory line was copied into the other, the two reports are not independent confirmations of the same count, and the agreement between them carries no evidential weight.
- Hypothesis: the strain count, not the phage count, is the discrepant term. The inconsistency could sit on the 188 side rather than on the phage width of the matrix.
Resolving Work
- Query the underlying susceptibility table directly in the KBase Data Lakehouse and count distinct phage identifiers, distinct strain identifiers, and non-null pair rows separately. Question: which of the three reported quantities, if any, does the stored data reproduce?
- Group the stored pair rows by phage and by strain and inspect the per-phage and per-strain row counts for ragged margins — that is, uneven row or column coverage where some phages or strains carry fewer calls than others. Question: is the matrix complete, or are missing cells the source of the pair count?
- Trace provenance of both report lines back to the upstream susceptibility experiment release and compare release dates and record counts. Question: did the phage inventory change between the versions the two projects read?
- Recompute every proportion each report derives from this matrix under each candidate denominator and record the range rather than a point estimate. Question: which downstream claims are denominator-sensitive and which survive all candidates?
- Check whether the per-family phage counts form a partition of the stated phage total in the stored taxonomy. Question: is the family breakdown consistent with either denominator?