BGPT: Create Tables: Extract Drug-Mechanism Matrix

Fuel Your Discoveries

Long Response Copied

Drug–Mechanism Matrix (extracted from provided full-text research extracts)

Matrix entries are restricted to explicit drug–mechanism or drug–target/MoA mappings present in the supplied research data. Where a study reports shared mechanism features across a drug class (e.g., HIV-1 PI resistance hotspots), the same mechanism tokens are applied to each listed drug only when the excerpt indicates that cross-drug relevance.

Included sources

(1) HIV-1 PI resistance ML benchmarking · (2) DeepTarget MoA (oncology) · (3) Corynebacterium image-based MoA antibiotics · (4) HPPD inhibitor chemistry/biology · (5) Sordarin & eEF2 diphthamide · (6) Yeast chemogenomics signatures (context only) · (7) Platinum resistance (context only) · (8) Melphalan ICL repair (MM) · (9) 5-HT2A PI hydrolysis & hallucinogens

Table A — Antibiotics vs MoA (Corynebacterium glutamicum image-based MoA study)

Drug → mechanistic target/process label as given in the extracted Table 1 list.

Drug	Mechanism / target category (as labeled)	Training in model (if stated)	MIC (µM) (if stated)
Amoxicillin	Cell wall (PBPs)	Included	0.342
Carbenicillin	Cell wall (PBPs)	Included	0.661
Cefotaxim	Cell wall (PBPs)	Included	0.263
Ampicillin	Cell wall (PBPs)	Not stated	0.358
Ethambutol	Cell wall (AG)	Included	4.895
Ciprofloxacin	DNA Gyrase	Included	0.755
Moxifloxacin	DNA Gyrase	Included	0.299
Novobiocin	DNA Gyrase	Included	26
Gepotidacin	DNA Gyrase	Not included	0.446
BDM71403	DNA Gyrase	Not included	0.268
Clarithromycin	Ribosome	Included	1.337
Doxycycline	Ribosome	Included	0.27
Linezolid	Ribosome	Included	2.964
Clofazimine	Cell membrane	Included	2.112
Rifampicin	RNA polymerase	Included	0.146
Rifabutin	RNA polymerase	Included	0.295
Trimethoprim	Folate synthesis	Not stated	>64
Sulfamethoxazole	Folate synthesis	Not stated	>64

Source basis for all drug→MoA/MIC rows above: extracted Table 1 drug labels and MICs from the Corynebacterium glutamicum imaging MoA paper .

Table B — Drug → Mechanism tokens in HIV-1 protease inhibitor resistance (from ML resistance benchmarking)

Mechanism tokens reflect resistance-associated protease hotspot positions reported by the excerpt; they are applied to each listed PI because the excerpt indicates hotspots are “key resistance” features for the PI-resistance setting.

HIV-1 Protease inhibitor (PI)	Resistance mechanism tokens (as hotspot positions)	Structural context token (if stated)
FPV (fosamprenavir)	Positions 10, 46, 54, 71, 90	Flap-region energetics highlighted (46–54)
IDV (indinavir)	Positions 10, 46, 54, 71, 90	Flap-region energetics highlighted (46–54)
NFV (nelfinavir)	Positions 10, 46, 54, 71, 90	Flap-region energetics highlighted (46–54)
SQV (saquinavir)	Positions 10, 46, 54, 71, 90	Flap-region energetics highlighted (46–54)
ATV (atazanavir)	Positions 10, 46, 54, 71, 90	Flap-region energetics highlighted (46–54)
LPV (lopinavir)	Positions 10, 46, 54, 71, 90	Flap-region energetics highlighted (46–54)
DRV (darunavir)	Positions 10, 46, 54, 71, 90	Flap-region energetics highlighted (46–54)
TPV (tipranavir)	Positions 10, 46, 54, 71, 90	Flap-region energetics highlighted (46–54)

Mechanistic hotspot + flap token: extracted from the HIV-1 PI resistance benchmarking paper’s results excerpt .

Critical note (uncertainty): the excerpt does not provide a per-PI position-importance matrix. Applying the same hotspot token set to each PI is therefore an assumption of shared resistance features within the PI-resistance modeling task; it is consistent with the excerpt phrasing but cannot be confirmed as PI-specific from the provided data.

Table C — Oncology drug → predicted (and some experimentally supported) mechanisms of action (DeepTarget)

Drug	Primary MoA / target (as described)	Context / secondary target token (as described)	Evidence type in excerpt
Pyrimethamine	Inhibition of mitochondrial oxidative phosphorylation (OXPHOS)	N/A (not stated in excerpt)	Chemogenomic CRISPR screen + pathway validation (case study)
Ibrutinib	Not specified as primary target in excerpt	Secondary target: EGFR in BTK-low contexts	Experimental validation in cell-line contexts (case study)

Case-study mappings are extracted from the DeepTarget excerpt .

Table D — Plant/animal HPPD inhibitors (reviewed mechanism tokens)

Compound / label	Target	Inhibition mode token (as described)	Activity / kinetics token (as stated)
NTBC / 14	HPPD (4-hydroxyphenylpyruvate dioxygenase)	β-triketone Fe(II)-center coordination; long-residence-time	Kd picomolar range; complex half-life ~63 h in rats, ~53 h in humans; rat liver HPPD IC50 ~40 nM
16	AtHPPD (Arabidopsis)	Plant-oriented HPPD inhibition; maize selectivity context	Ki not explicitly stated; noted maize selectivity tied to metabolism; PK t1/2 ~1 h in humans (per excerpt)
24	Daucus carota / PfHPPD (carrot; Pseudomonas fluorescens)	Isoxazole-derived cyanodiketone metabolite; binds Fe(II)-HPPD	Kd ~6 nM; complex half-life ~2 h; nearly irreversible in carrot system
4	AtHPPD	Pyrazole HPPD inhibitor	Ki ~12 nM toward AtHPPD
46	AtHPPD	Triketone-based quinazoline/quinazoline–hybrid	Ki = 5 nM (excerpt notes “better than 16”)
53 (usnic acid)	AtHPPD	Natural β-triketone-like scaffold	Ki = 70 nM
54 (pinocembrin)	Pig liver HPPD	Distinct binding mode token (non-classic vs triketones)	IC50 ~73 µM
28	AtHPPD	β-triketone with long alkyl (nonyl) substituent	Inhibitory activity ~19 nM

These compound-level mechanism tokens and activity/kinetics excerpts are taken directly from the HPPD inhibitor perspective’s extracted list .

Table E — Antifungal sordarin: eEF2 diphthamidation determinant of sensitivity

Drug	Mechanism token (ribosome/PTM determinant)	Evidence token in excerpt
Sordarin	Diphthamide modification of translation elongation factor 2 (eEF2) increases sordarin binding and toxicity	Yeast dphΔ mutants show increased resistance; cell-free sordarin binding reduced to ~60–80% vs WT in diphthamide-deficient extracts (normalized to eEF2); reduced [3H]sordarin binding in dphΔ extracts supports PTM-dependent binding

Derived from chemical-genetic screen excerpt .

Table F — Melphalan resistance in multiple myeloma: ICL repair after exposure

Drug / lesion type	Mechanism token	Observed in excerpt (naive vs treated)
Melphalan → DNA interstrand crosslinks (ICLs)	Enhanced repair of melphalan-induced ICLs at ~40 hours underlies clinical resistance in MM	ICL formation occurs in both naive and treated; no repair at 40 h in melphalan-naive patients (0%); treated patients show significant ICL repair at 40 h ranging ~42–100% (and sequential persistence in some patients)

Extracted from the melphalan resistance repair excerpt .

Table G — Hallucinogens vs 5-HT2A-stimulated phosphoinositide hydrolysis (necessity test)

Compound	5-HT2A → PI hydrolysis effect token	In vivo discrimination alignment (token from excerpt)
LSD	Concentration-dependent PI hydrolysis stimulation; 5-HT2A blocked by ketanserin	LSD generalizes “complete” to LSD/DOM training (excerpt)
DOM	Concentration-dependent PI hydrolysis stimulation; 5-HT2A blocked by ketanserin	DOM generalizes “complete” (excerpt)
psilocybin	PI hydrolysis stimulation	Generalizes “complete” (excerpt)
DMT	PI hydrolysis stimulation	Generalization “intermediate” (excerpt)
MDMT	PI hydrolysis stimulation	Generalization “intermediate” (excerpt)
DET	PI hydrolysis stimulation	Generalization “intermediate” (excerpt)
BOL	No PI hydrolysis stimulation (“NS”)	Not stated in excerpt
harmaline / harmane	No PI hydrolysis stimulation (“NS”)	Not stated in excerpt

Extracted from the 5-HT2A PI hydrolysis and discriminative stimulus excerpt .

Critical interpretation boundary: The excerpt itself concludes the PI hydrolysis pathway is not the sole mechanism for hallucinogen discriminative stimulus effects, and it discusses a possible threshold among multiple intracellular changes; therefore this table should be treated as a “signaling-token mapping,” not a definitive causal mechanism.

Figure 1 — Count of extracted antibiotics by MoA category (Table A)

Figure 2 — Mechanism-token coverage across extracted tables

A schematic “how many drugs/labels were explicitly mapped” per table section, based strictly on what the excerpt explicitly listed.

Blind spots & falsifiability checks (what could change the matrix)

PI hotspot token reuse: For HIV PIs, the excerpt provides hotspot position tokens but not a PI-by-PI mechanism matrix; Table B therefore cannot confirm PI-specificity.
Review perspective aggregation: HPPD Table D is based on a perspective review; values and mechanisms can vary by assay/system and may not be directly comparable across species without the full SI context .
Imaging phenotype mapping: Antibiotic MoA labels are mechanism-category tokens derived from the modeling task; they are not guaranteed to equal direct binding targets for every molecule outside the excerpt’s scope .

Feedback:

Updated: April 28, 2026

Top Data Sources Export MCP

1. A comprehensive benchmarking study showing that data set construction profoundly influences HIV-1 protease inhibitor resistance predictions, and that well-chosen physicochemical features (zScales) or structure-informed Rosetta terms with simple logistic regression can match complex neural networks, while highlighting distinct resistance mechanisms and the trade-offs between predictive performance and interpretability. [2025]

9QualityResults Limitations Context Blindspots Methods Sample Conflict Data

↗ Paper Review ↗ Full Paper

2. DeepTarget is a computational framework that integrates large-scale drug viability screens, CRISPR knockout dependencies, and omics data to predict cancer drug mechanisms of action—identifying primary targets, context-specific secondary targets, and target mutation preferences—and demonstrates strong performance across eight gold-standard benchmarks, with experimental validation in case studies (pyrimethamine-OXPHOS and Ibrutinib-EGFR) and links to open-source tools and resources. [2025]

9QualityResults Limitations Context Blindspots Methods Sample Conflict Data

↗ Paper Review ↗ Full Paper

3. Deep learning analyzes high-throughput images of chemically and genetically perturbed Corynebacterium glutamicum to predict antibiotic mechanisms of action, identify unseen MoAs, and link drug perturbations to pathway-level mutants for mutant-based target deconvolution. [2026]

8QualityResults Limitations Context Blindspots Methods Sample Conflict Data

↗ Paper Review ↗ Full Paper

4. This Perspective reviews 4-hydroxyphenylpyruvate dioxygenase (HPPD): its biological role in Tyr metabolism, the development and diversity of HPPD inhibitors (triketones, pyrazoles, isoxazoles), their applications as plant bleaching herbicides and potential therapeutic agents for human diseases (tyrosinemias, alkaptonuria, hawkinsinuria), and implications for drug repositioning, mechanism, pharmacokinetics, and safety. [2017]

9QualityResults Limitations Context Blindspots Methods Sample Data

↗ Paper Review ↗ Full Paper

5. A genome-wide chemical-genomic screen in Saccharomyces cerevisiae identifies diphthamide modification of translation elongation factor 2 (eEF2) as a key determinant of sordarin sensitivity, linking posttranslational modification to antifungal action and revealing new insights into the drug’s mechanism. [2008]

9QualityResults Limitations Context Blindspots Methods Sample Conflict Data

↗ Paper Review ↗ Full Paper

Keep Exploring

Which mechanisms are consistently predictive across different MoA inference modalities (sequence-based resistance, imaging phenotypes, chemogenomics, and structural/biochemical binding)?

Can we build a mechanistic ontology that aligns antibiotic targets, HIV protease resistance hotspots, kinase/EGFR context, and eEF2 PTM determinants into a single graph with evidence-strength scores?

Analysis Wizard

Build a unified drug–mechanism table by programmatically parsing the extracted paper snippets into structured rows, then export CSV/JSON and compute category counts for dashboard plots.

Science Art

Science Movie

Make a narrated HD Science movie for this answer ($32 per minute)

Discussion

BGPT Bias

I prioritize mechanistic mappings explicitly stated in the provided extracts, which can under-cover drugs/targets when studies imply but do not tabulate drug–mechanism pairs.

Get Ahead With Science Insights

Custom summaries of the latest cutting edge Science research. Every Friday. No Ads.

Built for bioinformatics workflows

Download raw tables, code snippets, and datasets from full texts to power analyses.

Fuel Your Discoveries

Long Response Copied

Drug–Mechanism Matrix (extracted from provided full-text research extracts)

Table A — Antibiotics vs MoA (Corynebacterium glutamicum image-based MoA study)

Table B — Drug → Mechanism tokens in HIV-1 protease inhibitor resistance (from ML resistance benchmarking)

Table C — Oncology drug → predicted (and some experimentally supported) mechanisms of action (DeepTarget)

Table D — Plant/animal HPPD inhibitors (reviewed mechanism tokens)

Table E — Antifungal sordarin: eEF2 diphthamidation determinant of sensitivity

Table F — Melphalan resistance in multiple myeloma: ICL repair after exposure

Table G — Hallucinogens vs 5-HT2A-stimulated phosphoinositide hydrolysis (necessity test)

Figure 1 — Count of extracted antibiotics by MoA category (Table A)

Figure 2 — Mechanism-token coverage across extracted tables

Blind spots & falsifiability checks (what could change the matrix)

Top Data Sources Export MCP

3. Deep learning analyzes high-throughput images of chemically and genetically perturbed Corynebacterium glutamicum to predict antibiotic mechanisms of action, identify unseen MoAs, and link drug perturbations to pathway-level mutants for mutant-based target deconvolution. [2026]

Ask a Follow-Up

Keep Exploring

Which mechanisms are consistently predictive across different MoA inference modalities (sequence-based resistance, imaging phenotypes, chemogenomics, and structural/biochemical binding)?

Can we build a mechanistic ontology that aligns antibiotic targets, HIV protease resistance hotspots, kinase/EGFR context, and eEF2 PTM determinants into a single graph with evidence-strength scores?

Analysis Wizard

Build a unified drug–mechanism table by programmatically parsing the extracted paper snippets into structured rows, then export CSV/JSON and compute category counts for dashboard plots.

Science Art

Science Movie

Make a narrated HD Science movie for this answer ($32 per minute)

Discussion

BGPT Bias

I prioritize mechanistic mappings explicitly stated in the provided extracts, which can under-cover drugs/targets when studies imply but do not tabulate drug–mechanism pairs.

Get Ahead With Science Insights

My BGPT

Trending

Built for bioinformatics workflows

Download raw tables, code snippets, and datasets from full texts to power analyses.

Fuel Your Discoveries

Long Response Copied

Drug–Mechanism Matrix (extracted from provided full-text research extracts)

Table A — Antibiotics vs MoA (Corynebacterium glutamicum image-based MoA study)

Table B — Drug → Mechanism tokens in HIV-1 protease inhibitor resistance (from ML resistance benchmarking)

Table C — Oncology drug → predicted (and some experimentally supported) mechanisms of action (DeepTarget)

Table D — Plant/animal HPPD inhibitors (reviewed mechanism tokens)

Table E — Antifungal sordarin: eEF2 diphthamidation determinant of sensitivity

Table F — Melphalan resistance in multiple myeloma: ICL repair after exposure

Table G — Hallucinogens vs 5-HT2A-stimulated phosphoinositide hydrolysis (necessity test)

Figure 1 — Count of extracted antibiotics by MoA category (Table A)

Figure 2 — Mechanism-token coverage across extracted tables

Blind spots & falsifiability checks (what could change the matrix)

Top Data Sources ExportMCP

3. Deep learning analyzes high-throughput images of chemically and genetically perturbed Corynebacterium glutamicum to predict antibiotic mechanisms of action, identify unseen MoAs, and link drug perturbations to pathway-level mutants for mutant-based target deconvolution. [2026]

Ask a Follow-Up

Keep Exploring

Which mechanisms are consistently predictive across different MoA inference modalities (sequence-based resistance, imaging phenotypes, chemogenomics, and structural/biochemical binding)?

Can we build a mechanistic ontology that aligns antibiotic targets, HIV protease resistance hotspots, kinase/EGFR context, and eEF2 PTM determinants into a single graph with evidence-strength scores?

Analysis Wizard

Build a unified drug–mechanism table by programmatically parsing the extracted paper snippets into structured rows, then export CSV/JSON and compute category counts for dashboard plots.

Science Art

Science Movie

Make a narrated HD Science movie for this answer ($32 per minute)

Discussion

BGPT Bias

I prioritize mechanistic mappings explicitly stated in the provided extracts, which can under-cover drugs/targets when studies imply but do not tabulate drug–mechanism pairs.

Get Ahead With Science Insights

My BGPT

Trending

Top Data Sources Export MCP