Per paper, how large is each one-variable contrast in reprogramming efficiency?

Answer

The corpus supports a collection of within-paper descriptive ratios, with explicit separation of intended single-variable comparisons, incompletely documented comparisons, and protocols that change several parameters. A ratio is dimensionless, but this alone does not make its arms comparable: denominators cancel only when they actually have the same definition. None of the supplied records establishes a delivery-class effect with both factor cocktail and handling matched and a pair-specific replicated test meeting the task's overturn criterion 50392590; 53358305; be5ed205; 52499370.

Summary

Warren's oxygen/factor design and Timilsina's post-transduction surface comparison provide interpretable intended single-variable rows. Lam's format contrast is a bundled protocol, while its factor-removal comparisons lack complete matching documentation. Manzini's Sendai comparisons are multi-parameter; its instrument contrasts need separate dose/input checks. Okur's shared scoring rule does not establish matched delivery protocols. Large ratios involving near-zero controls are retained with floor warnings rather than promoted to precise fold effects 50392590; be5ed205; 79dfe4d7; 476a1f3c; 93b589b0; 34f9fb83; 53358305.

Evidence

Ratios below are larger/smaller, calculated from the printed arm values unless explicitly labelled source-printed. A heading's cell, denominator, counting, uncertainty and matching information applies to every row in that group; row-specific exceptions are stated. “Not stated” means not established by the verified extraction, not necessarily absent from the paper. No ratio is pooled across papers.

Warren 2010, PMC3656821: intended single-variable comparisons

Cell type is dH1f human ES-derived fibroblasts for the oxygen/factor block. Both arms start with 50,000 cells and use a 1/6 replating fraction. The denominator is the average TRA-1-60-positive colony count per well divided by initially plated cells and scaled to the replated fraction; the same rule applies to both arms. The record also mentions TRA-1-81 confirmation. Scoring is day 18 for that block; n=6 per arm, colony counts mean ± SD, no named test for any of these pairs. Matching rests on Table 1's common experiment block and fixed conditions, rather than an explicit all-else-constant sentence. Floor warning: no 50392590.

| Parameter/class | Arm A, printed efficiency; colonies/well | Arm B, printed efficiency; colonies/well | Larger/smaller; larger arm Evidence locator | |---|---|---|---|---| | Oxygen, four factors; handling | 20% O2: 0.6%; 48 ± 18 | 5% O2: 2.7%; 228 ± 30 | 4.5; 5% O2 50392590, raw_data.one_variable_pairs[1] | | Oxygen, five factors; handling | 20% O2: 2.9%; 243 ± 42 | 5% O2: 4.4%; 367 ± 38 | about 1.52; 5% O2 50392590, raw_data.one_variable_pairs[2] | | Add LIN28 at ambient oxygen; factors | Four factors: 0.6%; 48 ± 18 | Five factors: 2.9%; 243 ± 42 | about 4.83; five factors 50392590, raw_data.one_variable_pairs[3] | | Add LIN28 at low oxygen; factors | Four factors: 2.7%; 228 ± 30 | Five factors: 4.4%; 367 ± 38 | about 1.63; five factors 50392590, raw_data.one_variable_pairs[4] |

All four rows: 50392590. The changing oxygen ratio across cocktails is descriptive interaction, not a tested interaction term.

The ROCK-inhibitor row uses BJ human fibroblasts and the five-factor cocktail, 300,000 cells and a 1/24 split, with the same denominator/numerator. The source Methods identify BJ as neonatal foreskin; the record's adult label is not used. Passage/precise donor age and fixed counting day are not established. Splitting is on day 7. n=6, colony counts mean ± SD; no pair-specific test. Table structure supports matching; an explicit all-else-constant sentence is not stated 50392590; https://doi.org/10.1016/j.stem.2010.08.012].

| Parameter/class | Arm A | Arm B | Ratio; larger | Floor | |---|---|---|---|---|---| | Y-27632; handling | Absent: 2.0%, 249 ± 21 colonies/well | Present: 2.6%, 326 ± 49 colonies/well | 1.3; present | No |

Source: 50392590.

Timilsina 2023 and Lin 2012: intended single-variable rows, with qualifications

| Paper/cell | Parameter/class and arms as printed | Ratio; larger | Denominator and numerator | Day, uncertainty/test and matching | Floor Evidence locator | |---|---|---|---|---|---|---| | Timilsina, PMC10525589; human gingival fibroblasts, donor age/passage not stated | Post-transduction surface, handling: W10% HSt-PMEDSAH-g 0.37% versus Matrigel 0.22% | 1.68; defined surface | Colonies divided by original 3×10^5 fibroblasts, same both arms; AP plus morphology | Day-28 classification described, exact efficiency-count day not explicitly tied; three independent replicates, unpaired t-test for this pair, p<0.0068; extracted Methods say mean±SD, while figure legend labels error bars SEM. Both arms come from shared transduction then split to surfaces; all-else-constant sentence not stated | No be5ed205, raw_data.one_variable_rows[0] | | Lin, PMC3348002; BJ neonatal foreskin fibroblasts, passage not stated | Add DNp73, factors: four-factor control 1.67 versus +DNp73 21.0 colonies | 12.57 calculated; 12.6 source-printed; +DNp73 | Per 10,000 parental cells seeded, same both arms; ES-like morphology, ALP shown in legend | Day 21 both; count replicates, dispersion and test not stated. Both replated on Matrigel; all-else-constant sentence not stated | Yes: control approximately one or two colonies be5ed205, raw_data.one_variable_rows[1] |

Both rows: be5ed205. Timilsina tests surface after shared transduction, not different surfaces during initial viral exposure. The SD/SEM distinction is retained as a reporting ambiguity, not silently resolved.

Lam 2022, PMC9357361: factor-removal candidates, matching incomplete

All rows count TRA-1-60-positive colonies at day 14 per initial cell seeding. Cells are HFF-01 human foreskin fibroblasts or PBMC; donor age/passage not established. n=3, mean ± SD. The table's asterisks compare formats, not four versus three factors. Thus no test for these factor-removal pairs is established. The three-factor protocol and remaining-vector doses were not read, so all-else-constant is not stated; these remain candidates, not certified single-variable interventions 79dfe4d7.

| Cell/format | Parameter/class | Four-factor arm, percent mean±SD | Three-factor arm, percent mean±SD | Calculated ratio; larger | Floor | |---|---|---|---|---|---| | HFF-01 monolayer | Omit c-MYC; factors | 0.04 ± 0.02 | 0.0009 ± 0.004 | about 44; four factors | Yes: three-factor colonies 3 ± 1; reported SD exceeds mean | | HFF-01 microcarrier | Omit c-MYC; factors | 0.97 ± 0.01 | 0.15 ± 0.03 | about 6.5; four factors | No | | PBMC monolayer | Omit c-MYC; factors | 0.02 ± 0.01 | 0.0006 ± 0.0002 | about 33; four factors | Yes: three-factor colonies 2 ± 1 | | PBMC microcarrier | Omit c-MYC; factors | 0.97 ± 0.1 | 0.11 ± 0.01 | about 8.8; four factors | No |

All rows: 79dfe4d7. The quoted dispersions are the supplied record's values; no uncertainty is calculated for these ratios.

Manzini 2015, PMC4653244: instrument candidates and multi-parameter method contrasts

Cells are primary patient dermal fibroblasts hDF107, hDF137 and hDF119; donor age and starting passage are not established. Efficiency is “nascent colonies per cells transduced/transfected”, with morphology-based counting rather than TRA-1-60 counting of established lines. Scoring day is not stated; picking dates are not counting dates. Three independent rounds per method are described, but pair-specific dispersion and tests are not established. Both electroporation systems use the same plasmid cocktail and reported DNA dose; a universal all-else-constant sentence is not stated. Methods cell inputs differ from Figure 2's accounting, so the cell-number inconsistency remains unresolved cc55abeb; 269004f2; 93b589b0; e8f00280.

The following table preserves all twelve reported method pairs once. S=Sendai, N=Nucleofector, E=Neon, L=Lipofectamine. Percentages are printed values. Instrument rows are within-class candidates; electrical programmes differ as part of the instrument intervention. S rows additionally change delivery class, proprietary vector/factor preparation, input cells and dose conventions. L versus electroporation also changes delivery technique, DNA amount and input/confluence; it is not a strict one-variable comparison merely because plasmids are shared 476a1f3c; 93b589b0.

| Donor | Arm A, % | Arm B, % | Ratio; larger | Classification | Floor | |---|---|---|---|---|---| | 107 | S 0.0026 | N 0.017 | 6.54; N | Multi-parameter | No | | 107 | S 0.0026 | E 2.7e-5 | 96.3; S | Multi-parameter | Yes | | 107 | N 0.017 | E 2.7e-5 | about 629.6; N | Instrument candidate | Yes | | 137 | S 0.04 | N 0.0166 | 2.41; S | Multi-parameter | No | | 137 | S 0.04 | E 0.0044 | 9.09; S | Multi-parameter | No | | 137 | S 0.04 | L 0.001 | 40; S | Multi-parameter | Borderline low L | | 137 | N 0.0166 | E 0.0044 | 3.77; N | Instrument candidate | No | | 137 | N 0.0166 | L 0.001 | 16.6; N | Technique plus dose/input | Borderline low L | | 137 | E 0.0044 | L 0.001 | 4.4; E | Technique plus dose/input | Borderline low L | | 119 | S 0.0173 | N 0.003 | 5.77; S | Multi-parameter | No | | 119 | S 0.0173 | E 5.5e-5 | about 314.5; S | Multi-parameter | Yes | | 119 | N 0.003 | E 5.5e-5 | about 54.5; N | Instrument candidate | Yes |

Rows: 476a1f3c; 50392590. A correction record properly moves seven Sendai pairs out of the earlier single-variable list; its remaining five within-class labels still do not erase the two lipofection dose/input confounds 476a1f3c; 93b589b0. Lipofection produced no colonies in the other donors: retain that outcome without division by zero cc55abeb.

Two additional donor/cell-line descriptive rows keep method fixed. Denominator, counting/day and statistics are as above, with one patient line per arm and no matched-donor causal inference:

| Parameter/class | Arm A, % | Arm B, % | Ratio; larger | Floor | |---|---|---|---|---| | Donor within Sendai | hDF137 0.04 | hDF107 0.0026 | about 15.4; 137 | No | | Donor within Nucleofector | hDF107 0.017 | hDF119 0.003 | about 5.7; 107 | No |

Source: cc55abeb; 269004f2.

Simara 2014, PMID 25369347: surface-associated protocol packages

The verified record calls these surface-only rows. The primary Methods, however, specify different medium on MEFs and Geltrex, so shared medium is not established and these are conservatively placed among multi-parameter comparisons. Cells are human dermal fibroblasts or human CD34+ progenitors; the record does not establish donor age/passage. Denominator: input 10^5 cells, same both arms. Numerator: proper hESC-like colonies. Scoring day per arm, replicates, dispersion and tests are not stated; picking over days 17–28 is a window, not a counting date. Matching is supported for named method and input, not for all culture conditions 50392590; https://doi.org/10.14712/fb2014060S10085].

| Cell/method | Arm A: MEF colony count | Arm B: Geltrex colony count | Ratio; larger | Parameter class | Floor | |---|---|---|---|---|---| | Fibroblasts/Sendai | 7 | 21 | 3.0; Geltrex | Surface plus medium; handling package | No | | Fibroblasts/episomal | 10 | 27 | 2.7; Geltrex | Surface plus medium; handling package | No | | CD34+/Sendai | 5 | 16 | 3.2; Geltrex | Surface plus medium; handling package | Low count; no detection limit stated | | CD34+/episomal | 14 | 50 | about 3.57; Geltrex | Surface plus medium; handling package | No |

Counts/ratios: 50392590. These counts are retained; a stronger causal classification is not.

Lam 2022: multi-parameter format comparisons

All use the same Sendai delivery class, TRA-1-60 day-14 colonies per initial cell seeding, n=3, mean±SD. HFF-01 is human foreskin and IMR90 fetal lung fibroblasts; PBMC, CD3+ and CD34+ blood fractions are separately retained. Donor ages/passages are not stated. Format, agitation, substrate, medium/hydrogel and seeding procedures co-vary; fibroblast transduction efficiency also changes. All-else-constant: not stated. Table asterisks compare formats, not fractions or factor sets. HFF-01 format efficiency is marked p<0.0001; no unprinted exact p-value is assigned to the other pairs 79dfe4d7; ae2e9309; 50392590.

| Cell/factor set | Monolayer, percent mean±SD | Microcarrier, percent mean±SD | Source-printed efficiency fold | Ratio of rounded means, calculated | Floor | |---|---|---|---|---|---| | HFF-01, four factors | 0.04 ± 0.02 | 0.97 ± 0.01 | 24.5 ± 5.9 | 24.25 | No | | IMR90, four factors | 0.03 ± 0.01 | 0.61 ± 0.04 | 18.0 ± 2.2 | about 20.33 | No | | PBMC, four factors | 0.02 ± 0.01 | 0.97 ± 0.1 | 47.6 ± 7.5 | 48.5 | No | | CD3+, four factors | 0.02 ± 0.001 | 0.85 ± 0.05 | 41.3 ± 3.7 | 42.5 | No | | CD34+, four factors | 0.02 ± 0.01 | 0.5 ± 0.1 | 36.4 ± 9.2 | 25 | No | | HFF-01, three factors | 0.0009 ± 0.004 | 0.15 ± 0.03 | 184.7 ± 43.5 | about 166.67 | Yes | | PBMC, three factors | 0.0006 ± 0.0002 | 0.11 ± 0.01 | 187 ± 59.9 | about 183.33 | Yes |

All rows: 79dfe4d7. Microcarrier is larger in every row. The reported fold statistic is retained separately from arithmetic on rounded means; neither is substituted for the other. HFF-01 colony totals have a different ratio from efficiencies, so colony-count ratios cannot replace input-normalized efficiency ratios. The isolated agitation comparison is qualitative in the main text, with numeric values in unread Table S1 ae2e9309.

Other multi-parameter delivery and protocol comparisons

Warren: dH1f fibroblasts, KMOS modified mRNA 1.4% versus retrovirus 0.04%; calculated ratio 35, authors' stated ratio 36, RNA larger. Table colony counts are 229±39 versus 13±3.5 per well; n=18 RNA wells and n=9 viral wells, no pair-specific test established. Same input-scaled TRA-1-60 rule, but split fractions differ, and RNA ends on day 16 whereas viral cultures are fixed on day 32. Oxygen, feeders/medium and delivery also differ. Floor: no. Only cell line, factor names, initial plating and split day are shared; not all else 4bf6a2bd; 50392590.

Okur, PMC6637500: patient BM-MSCs are early passage P3–P4; control donor passage/age is not established. AP-positive colonies divided by all cells used; day 24–28 window for both, exact per-arm day not stated. The patient contrast has mean±SEM in the later verified record; actual experimental replicate count and pair-specific test are not established. Three patients is not automatically three experimental repeats. Control episomal spread is absent. Shared culture conditions/scoring do not match factor composition, delivery input or replating, so the earlier clean-pair label is not accepted 34f9fb83; 53358305.

| Cell group | SeV efficiency | Epi5 efficiency | Ratio; larger | Floor | |---|---|---|---|---| | Osteopetrotic patient MSC | 0.3 ± 0.11% | 0.1 ± 0.02% | 3.0; SeV | No | | Healthy-control MSC | 0.14 ± 0.01% | 0.07%, spread not stated | 2.0; SeV | No |

Rows: 34f9fb83; 53358305.

Trevisan, PMC5297836: neonatal foreskin BJ fibroblasts, passage 3; TRA-1-60 colonies over starting cells, also described as transduced cells. Approximate efficiencies: SeV 0.05%, episomal 0.05%, retrovirus 0.01%. SeV/retrovirus and episomal/retrovirus each give about 5; SeV/episomal about 1. No floor identified. Scoring day not stated. The verified extraction did not establish replicates or a pair-specific test; primary Figure 2 does contain replication and significance annotations, so that extraction omission must not be read as proof the article reports none. Those additional statistics are not used as verified deciding evidence here. Delivery, factor/vector composition, input, dose, ROCK inhibitor and replating co-vary despite shared starting passage/operator/material batches 98344c43; https://doi.org/10.3390/ijms18010206].

Goh, PMC3841145: BJ neonatal foreskin fibroblasts, early passage below 12 where established; TRA-1-81 colony counting. All-else-constant is not stated. These rows remain descriptive because surfaces, cocktails, cell inputs, dose, endpoint or denominator matching is incomplete be5ed205.

| Pair/parameter | Printed arm A | Printed arm B | Calculated ratio; larger | Denominator/day; pair-specific statistics | Floor/covariates | |---|---|---|---|---|---| | Episomal surface/package | Matrigel 0.103 ± 0.008% | Feeders 0.012 ± 0.008% | 8.58; Matrigel | Transfected-cell correction; general day-30 rule, individual-arm day not restated; n=3 Matrigel, feeder n not established for this exact pair; no test | Yes, broad spread in feeder arm; medium, reseeding, dish also differ | | Episomal vector/factor package | Yamanaka 0.012 ± 0.008% | Thomson 0.003 ± 0.001% | 4.0; Yamanaka | Corrected transfected cells; day 25–30 noted for Yamanaka, other day not separately stated; n=2 each, no test | Low-count/spread warning; factors, backbone, DNA dose and correction differ | | mRNA surface/package | Feeders 1.89% | Matrigel 0.22% | about 8.59; feeders | Starting-cell denominator/day 18 for feeders; Matrigel denominator/day not established; n/spread/test not stated | No floor identified; surface differs, endpoint matching unknown | | Delivery package | mRNA 1.89% | Retrovirus 0.2% | 9.45; mRNA | Starting versus estimated transduced cells; day 18 versus 30; pair-specific n/spread/test not stated | No floor identified; delivery, scoring, denominator and handling differ | | Delivery package | mRNA 1.89% | Episomal 0.10% | 18.9; mRNA | Starting versus corrected transfected cells; day 18 versus 30; pair-specific n/spread/test not stated | No floor identified; delivery, surface, scoring and denominator differ |

All rows: be5ed205. The text's “over 10 fold” vector claim is not reproduced by its printed efficiencies; no ratio from that prose is substituted for the arithmetic. Uncertainty in denominator matching remains even inside one paper.

Kuebler, PMC12834790: primary skin fibroblasts from unpaired donors, passage 3–12, cohort ages spanning 0–107 years, individual-arm ages not established. Emerging-colony counts per somatic cells seeded; exact scoring day and marker-specific counting rule not stated. mRNA 0.221% (SD 0.235) versus episomal 0.017% (SD 0.027) gives 13 by means; median ratio is separately reported as 31.8. Fibroblast SeV 0.076% versus episomal 0.017% gives about 4.5. These change factor cocktails, cell input, culture/replating and donor assignment; RNA oxygen is specified while episomal oxygen is unknown, not proven different. All-else-constant is not stated. No source-established detection floor is assigned 52499370.

For mRNA versus episomal, the reported adjusted GLM/Tukey p=0.0412 applies to a model contrast rather than the raw ratio of means; subset n is not established. Episomal versus SeV has reported p=0.0116 for that model contrast. The early/late passage model result 0.704 (95% CI 0.303–1.64, p=0.4087) is not a pair of raw efficiencies and is therefore retained as a non-qualifying model contrast, not converted into a new larger/smaller ratio 52499370.

Mackey 2018, PMC6696928: observational donor-group rows

Human adult dermal fibroblasts, passage below 10 (mostly 1–5), one lentiviral six-factor protocol. Numerator is AP-stained colonies counted by ImageJ; denominator is colonies/250,000 input cells, the same rule for both arms. Culture lasts 21 days in E8 before colonies appear; exact AP scoring date is not separately established. Each donor line is run in triplicate; most have low within-line variation, but group-mean dispersion is not stated. Matching all other donor properties is not stated. These are donor associations, not randomized single-cause tests bec4bd8c.

| Parameter/class | Arm A | Arm B | Ratio; larger | Pair-specific test | Floor | |---|---|---|---|---|---| | Self-reported ancestry; donor | AA n=41, 1634 colonies | EA n=39, 1175 colonies | 1.39; AA group | Spearman association, p=0.01 | No for group means | | Age group; donor | ≤33 years n=41, 1601 colonies | >33 years n=39, 1221 colonies | 1.31; younger group | Spearman association, p=0.03 | No for group means |

Source: bec4bd8c. The separate ancestry and age associations do not establish independent causal effects; age-group findings are not generalized to a continuous-age relationship.

Hasegawa 2010, PMID 20572011: route comparison reported as bounds

Human fibroblasts, donor age and passage not stated in the accessible abstract. Factor-transduction iPSC generation is reported at <0.001%, versus >0.005% for hESC–fibroblast fusion. The supplied record conservatively describes a lower-bound advantage of at least about five for fusion; this is not a measured fivefold effect or an exact ratio. With a positive nonzero factor-transduction efficiency, the printed inequalities imply a ratio greater than five; an actual zero would make division undefined. Floor warning: yes, the smaller arm has only an upper bound 17f58a64.

Parameter/class: reprogramming route or mechanism. This remains a multi-parameter comparison: route and completion timeline differ. Durations are >4 weeks versus <10 days, not fixed scoring days. Numerator, denominator, denominator matching, per-arm scoring date, replicates, dispersion, statistical test and an all-else-constant statement are not established by the abstract. The primary abstract was checked; the publisher PDF route returned an access refusal. Missing details are therefore access-limited, not absent from the full paper. No table-derived point values were imported 17f58a64.

Methodology

All sixteen verified records are accounted for inline. Repeated extractions of Lam and Manzini are reconciled into single tables rather than counted as independent experiments. Source checking read the available primary Methods, Results, table text and relevant legends for Lam, Manzini, Okur, Trevisan, Timilsina, Lin, Goh and Kuebler; Warren and Mackey were checked through platform primary-text retrieval after the XML route failed. Simara's publisher PDF was reached and its culture Methods checked. The added Hasegawa record was checked against the primary abstract; no full-text table was reached. No new literature search was used to claim completeness; the non-seed additions in the supplied search record are Timilsina, Lin and Goh. Negative candidate reads and unread supplements remain access-limited observations, not assertions that a study has no such data be5ed205; 98344c43; 34f9fb83.

Ratios of rounded means, source-reported fold statistics and adjusted model estimates are kept distinct. No ratios or efficiencies are averaged across papers. No structured measurement conflicts were supplied. Disputed clean-versus-multiple-variable labels are methodological classification issues, not incompatible measurements of the same biological quantity. Primary checks are used to limit overstatement of audited extractions, not to import unverified new numerical findings.

Limitations

Missing all-else-constant statements remain visible. A table's common denominator formula is weaker than proof of equal experimental input or equal transfection-success correction. Near-zero controls and uncertainty of rounded means limit ratios. Unread Lam supplementary isolation/three-factor protocols and agitation values, unextracted source figures and ambiguous scoring days prevent a complete census. The corpus contains no matched, replicated delivery-class contrast with a pair-specific test satisfying the overturn criterion; that is a bounded corpus result, not a claim that no such study exists 79dfe4d7; ae2e9309; 50392590; be5ed205; 52499370.

Response to the refusal: traceability of the figures

The quoted oxygen row is explicitly present in the supplied VERIFIED corpus, not inferred from an abstract. In submission 50392590, raw_data.one_variable_pairs[1].arm_A gives the ambient-oxygen efficiency and colony count, arm_B gives the low-oxygen values, and ratio supplies their quotient; the same object supplies denominator, day, replication, dispersion and matching. The next three objects contain the other oxygen/factor rows; object [0] contains the ROCK-inhibitor row. The Warren table now places these locators directly beside each row. The independent Warren extraction also repeats the four-factor oxygen values and five-factor efficiencies in raw_data.finding 50392590; 4bf6a2bd.

The Timilsina and Lin rows are similarly explicit in submission be5ed205, raw_data.one_variable_rows[0] and [1], respectively: arm_A, arm_B and ratio contain the numbers; the denominator, scoring_day, replicates_dispersion_test and everything_else_held_constant fields contain the methodological qualifications. Primary checking confirms Timilsina's efficiencies in Results beside Figure 5, with the replicate/error-bar information in its legend and the shared-transduction procedure in Methods; Lin's Results subsection “DNp73 overexpression enhanced human iPS cell generation” explicitly prints the two colony counts and the rounded enhancement. Their limitations remain in the report, including the Timilsina SD/SEM ambiguity and Lin's near-floor control be5ed205.

The objection's general reference to other ratios does not identify additional rows, so the other numerical groups were also checked: Lam's full arm values and printed fold statistics are in raw_data.one_variable_pairs and multi_parameter_pairs; Manzini's corrected method-pair set is in the correction record alongside the instrument/dose/input disclosures; Goh's rows are in raw_data.multi_parameter_rows of the same multi-paper record used for Timilsina and Lin. Kuebler's arm values and model-test distinctions and Mackey's donor-group counts are explicit in their respective raw_data fields. These remain separate within-paper calculations, with no pooled ratio, fabricated statistical test or erased confound 79dfe4d7; 476a1f3c; 93b589b0; be5ed205; 52499370; bec4bd8c.