BoltzMaker Report | Marc C. Deller

Campaign summary

Field	Value	Details
Input file	boltz_input.md	boltz_input.md
Proteins	1	2 protein block(s) in 1 group(s): ABL1 (1130 aa); ABLAP (1130 aa)
Co-folded partners	0	none
Ligands	2	IMATI (SMILES); DASAT (SMILES)
Pockets	1	1N1
Apo structure references	1	ABLAP
Predictions	5	ABL1_IMATI, ABL1_IMATI_1N1, ABL1_DASAT, ABL1_DASAT_1N1, ABLAP
Predict affinity	yes	pIC50 predicted for every ligand-bound prediction
Ligand chemistry	2 of 2 flagged	IMATI, DASAT -- see "Ligand preparation" below
Boltz predict runtime	2h 33m 26s	across 2 run invocations
Accelerator	gpu	gpu = Metal/CUDA backend used; cpu = no GPU available
Workers	0	parallel data-loading workers (Boltz's own default is 2)
MPS watermark	1.0	PYTORCH_MPS_HIGH_WATERMARK_RATIO cap -- lower avoids swap on Apple unified memory
Max parallel samples	1	Boltz's own --max_parallel_samples
Max MSA sequences	4096	cap on MSA sequences used for co-evolution features

Reference structures

The experimental structures this campaign was built on: where each pocket came from, and what the secondary-structure comparison measures against.

Pocket definitions

Pocket	From	Proteins	Contacts
1N1	2GQG	ABL1	18
STI	1IEP	ABL1	reference only

Secondary-structure references

Protein	Structure	State	Contents	Chain
ABL1	2gqg.cif	DFG-in, αC-in	2 chain(s), bound: 1N1, PTR	auto

A kinase's state is its DFG motif and αC helix, taken from the secondary-structure comparison.

Ligand definitions

Ligand	Class	Given as	Experimental structure
IMATI	Control	SMILES	STI
DASAT	Control	SMILES	1N1

A control has been verified experimentally, by structure or by assay, so its prediction can be checked; an experimental compound is under investigation, with nothing to check against. Experimental compounds are ringed in red in the charts.

Summary table

Identity						Affinity		▸ Confidence (5)	Structure
Run	Protein	Ligand	Pocket	Class	Summary	Binder p	pIC50	Score	pTM	ipTM	Lig ipTM	pLDDT	CIF
4	ABL1	DASAT	1N1	Control	⚠️	0.99	12.41	0.64	0.54	0.98	0.98	0.55	CIF
2	ABL1	IMATI	1N1	Control	⚠️	0.83	10.38	0.63	0.54	0.98	0.98	0.55	CIF
3	ABL1	DASAT	Unc	Control		0.92	10.33	0.53	0.41	0.84	0.84	0.46	CIF
1	ABL1	IMATI	Unc	Control		0.36	8.33	0.47	0.40	0.57	0.57	0.45	CIF
5	ABLAP	Apo	N/A	N/A	N/A	N/A	N/A	0.53	0.48	N/A	N/A	0.54	CIF

Download full CSV · Download summary CSV

🎯 affinity · 🛡️ confidence: Likely binder (p ≥ 0.7) Uncertain (0.3-0.7) Likely non-binder (p < 0.3) High confidence (≥ 0.7) Moderate confidence (0.5-0.7) Low confidence (< 0.5)

Pockets

1 named pocket(s) (1N1) plus an unconstrained baseline; contacts are enforced within 4 A.

Pocket	Protein	Ligands	Runs	Targets	Contacts
1N1	ABL1	IMATI, DASAT	2, 4	2	18 residue(s)
Unconstrained	ABL1	IMATI, DASAT	1, 3	2	none -- ligand placed freely

Ligand pose vs experiment

4 target(s) compared against experimental structures in reference/. Atoms are paired by molecular graph, so the symmetry of the ligand is respected rather than being resolved by whichever atom happened to be nearest.

Target	Protein	Ligand	Pocket	Reference	Site (A)	Pose (A)	Conformer (A)
ABL1_IMATI_1N1	ABL1	IMATI	1N1	1IEP (STI)	0.38	0.56	0.25
ABL1_DASAT_1N1	ABL1	DASAT	1N1	2GQG (1N1)	0.97	2.03	1.73
ABL1_DASAT	ABL1	DASAT	unconstrained	2GQG (1N1)	1.48	2.66	1.92
ABL1_IMATI	ABL1	IMATI	unconstrained	1IEP (STI)	14.66	18.39	2.48

🎯 site · pose: Same site (< 2 A) Nearby (2-5 A) Elsewhere (≥ 5 A) Pose reproduced (< 2 A) Approximate (2-5 A) Different pose (≥ 5 A)

Predicted against experimental

4 pair(s) across 2 pocket(s). Each frame holds exactly two ligands, superposed through their receptors: the prediction in red, the experimental one in grey.

Pocket 1N1

ABL1_IMATI_1N10.56 Å

Grey: 1IEP STI. Red: predicted.

ABL1_DASAT_1N12.03 Å

Grey: 2GQG 1N1. Red: predicted.

Unconstrained

ABL1_DASAT2.66 Å

Grey: 2GQG 1N1. Red: predicted.

ABL1_IMATI18.39 Å

Grey: 1IEP STI. Red: predicted.

Ligand preparation

2 of 2 ligand(s) flagged for chemistry review -- these are advisory, not errors; verify the input SMILES reflects what you intended before trusting the results below.

Ligand	Chemistry notes
IMATI	ionizable group(s) present (primary/secondary amine) -- verify the SMILES reflects your intended protonation state
DASAT	ionizable group(s) present (primary/secondary amine) -- verify the SMILES reflects your intended protonation state

Ligand structures

No shared scaffold or substructure detected across the set -- ligands are structurally distinct.

Sundefined stereocentreAcarboxylic acidNprimary/secondary aminePhphenolSO3sulfonic acidsaltsalt/disconnected fragment

IMATIN

MW 494 · cLogP 4.6 · TPSA 86

DASATN

MW 488 · cLogP 3.3 · TPSA 107

Download PDF · Download SMILES

Scaffolds: Bemis-Murcko, exact match first, then Tanimoto-clustered (Morgan r=2, 2048-bit, threshold 0.60) whole-group MCS as a verified fallback. Minimum highlighted substructure size: 8 heavy atoms. Stereocentre/ionizable-group highlighting from this campaign's own ligand-preparation check (see above).

pIC50 vs confidence score

pIC50 vs binder probability

Ranked predicted pIC50

Ranked confidence

ABL1: residue interaction fingerprint

Family coverage and SSE shifts

Family	Status	Detail
ABL1	OK	28 motif row(s) across 4 target(s), annotator=kinase
ABLAP	No apo structure configured	No 'Apo structure:' configured for this family

2 target(s), 28 motif(s) compared
Mean Ca RMSD: 1.64 A (median 1.55 A) — largest shift: 5.36 A at ABL1_IMATI / DFG
Mean centroid shift: 1.14 A
Flagged phi/psi outlier residues: 57
Kinase state changes detected: 28 DFG, 7 alphaC

Identity						Shift			Backbone	Kinase state
Family	Target	Ligand	Motif	Kind	Source	N res	RMSD (A)	Centroid delta (A)	Flagged phi/psi	DFG apo	DFG holo	DFG delta	alphaC apo	alphaC holo	alphaC delta
ABL1	ABL1_DASAT	DASAT	DFG	loop	kinase	3	2.91	1.22	3	in	out	True	in	in	False
ABL1	ABL1_DASAT	DASAT	DFG	loop	kinase	3	3.23	1.70	3	in	other	True	in	in	False
ABL1	ABL1_DASAT	DASAT	catalytic_loop	loop	kinase	3	1.63	1.63	0	in	out	True	in	in	False
ABL1	ABL1_DASAT	DASAT	catalytic_loop	loop	kinase	3	0.91	0.90	0	in	other	True	in	in	False
ABL1	ABL1_DASAT	DASAT	hinge	loop	kinase	3	0.41	0.37	0	in	out	True	in	in	False
ABL1	ABL1_DASAT	DASAT	hinge	loop	kinase	3	0.49	0.45	0	in	other	True	in	in	False
ABL1	ABL1_DASAT	DASAT	pocket_scaffold	loop	kinase	73	1.67	0.00	12	in	out	True	in	in	False
ABL1	ABL1_DASAT	DASAT	pocket_scaffold	loop	kinase	73	1.59	0.00	10	in	other	True	in	in	False
ABL1	ABL1_DASAT	DASAT	alphaC_Glu	point	kinase	1	2.07	2.07	0	in	out	True	in	in	False
ABL1	ABL1_DASAT	DASAT	alphaC_Glu	point	kinase	1	1.87	1.87	0	in	other	True	in	in	False
ABL1	ABL1_DASAT	DASAT	catalytic_Lys	point	kinase	1	1.33	1.33	0	in	out	True	in	in	False
ABL1	ABL1_DASAT	DASAT	catalytic_Lys	point	kinase	1	0.67	0.67	0	in	other	True	in	in	False
ABL1	ABL1_DASAT	DASAT	gatekeeper	point	kinase	1	0.36	0.36	0	in	out	True	in	in	False
ABL1	ABL1_DASAT	DASAT	gatekeeper	point	kinase	1	0.55	0.55	0	in	other	True	in	in	False
ABL1	ABL1_IMATI	IMATI	DFG	loop	kinase	3	3.22	2.08	3	in	other	True	in	out	True
ABL1	ABL1_IMATI	IMATI	DFG	loop	kinase	3	5.36	4.23	2	in	other	True	in	in	False
ABL1	ABL1_IMATI	IMATI	catalytic_loop	loop	kinase	3	2.45	2.43	0	in	other	True	in	out	True
ABL1	ABL1_IMATI	IMATI	catalytic_loop	loop	kinase	3	0.40	0.30	0	in	other	True	in	in	False
ABL1	ABL1_IMATI	IMATI	hinge	loop	kinase	3	0.47	0.29	0	in	other	True	in	out	True
ABL1	ABL1_IMATI	IMATI	hinge	loop	kinase	3	0.59	0.55	0	in	other	True	in	in	False
ABL1	ABL1_IMATI	IMATI	pocket_scaffold	loop	kinase	73	2.13	0.00	12	in	other	True	in	out	True
ABL1	ABL1_IMATI	IMATI	pocket_scaffold	loop	kinase	73	2.73	0.00	12	in	other	True	in	in	False
ABL1	ABL1_IMATI	IMATI	alphaC_Glu	point	kinase	1	3.44	3.44	0	in	other	True	in	out	True
ABL1	ABL1_IMATI	IMATI	alphaC_Glu	point	kinase	1	1.84	1.84	0	in	other	True	in	in	False
ABL1	ABL1_IMATI	IMATI	catalytic_Lys	point	kinase	1	1.32	1.32	0	in	other	True	in	out	True
ABL1	ABL1_IMATI	IMATI	catalytic_Lys	point	kinase	1	1.51	1.51	0	in	other	True	in	in	False
ABL1	ABL1_IMATI	IMATI	gatekeeper	point	kinase	1	0.46	0.46	0	in	other	True	in	out	True
ABL1	ABL1_IMATI	IMATI	gatekeeper	point	kinase	1	0.39	0.39	0	in	other	True	in	in	False

Download CSV

Per-motif Ca RMSD

Selectivity and motif shifts

Family x ligand selectivity

Motif x target RMSD

Landlord narration

4 of 5 target summaries written on-device by the Apple Neural Engine; 1 rendered from the template. Every number here is checked against the figures the analysis computed; a summary stating one it was not given was replaced by the template.

The ABL_KINASE campaign tested the ability to predict the structure of ABL1 and ABLAP, using DASAT and IMATI as ligands. Of 5 predicted targets, 4 were flagged as caution and 1 as discard.

Targets	5 predicted, 4 with a ligand
Receptors	ABL1, ABLAP
Ligands	DASAT, IMATI
Verdicts	4 caution, 1 discard
Confidence	3 poorly determined, 2 moderately determined
Flagged	5 of 5
Pose validated	1 of 4 reproduced the experimental pose

Key findings

Of 5 targets, 0 are marked proceed, 4 caution and 1 discard.
0 of 5 targets are well determined; 5 are not.
5 of 5 targets carry at least one flag.
4 targets could be checked against an experimental structure, and 1 reproduced the experimental pose.

Highest predicted potency

DASAT on 4_ABL1_DASAT_1N1, predicted pIC50 12.41
IMATI on 2_ABL1_IMATI_1N1, predicted pIC50 10.38
DASAT on 3_ABL1_DASAT_Unc, predicted pIC50 10.33

Caveats

The campaign was limited by the fact that the confidence of the structure was moderately determined, meaning that it is somewhat reliable but could still be improved.

Per target

Target	Verdict	Summary	Caveat	Written by
`ABL1_IMATI`	discard	The structure is poorly determined, and this means that someone deciding whether to trust it should be cautious. IMATI is poorly placed relative to the receptor and has a predicted potency of 8.33, ranking it fourth among four ligands tested	Overall confidence is below the well-determined threshold.	model
`ABL1_IMATI_1N1`	caution	The structure is moderately determined, meaning it is likely to be accurate, but not absolutely so. This level of certainty may be acceptable for some applications, but not for others where high precision is required. IMATI is a control ligand with low predicted potency, indicating that it is unlikely to bind to ABL1 effectively. Its interaction analysis has not been run, which may affect the accuracy of the prediction.	Overall confidence is below the well-determined threshold, meaning the structure is not considered to be reliable enough for most applications.	model
`ABL1_DASAT`	caution	The structure is poorly determined: confidence 0.534, ipTM 0.835, ligand ipTM 0.835. DASAT (control) at Unc: predicted pIC50 10.33, binder probability 0.922, ranked 3 of 4 by predicted potency. Interactions: interaction analysis not run. Against 2GQG the prediction close to the experimental pose (pose 2.66 Å).	overall confidence is below the well-determined threshold.	template
`ABL1_DASAT_1N1`	caution	The confidence of this structure is moderately determined, meaning that it is somewhat reliable but could still be improved. This suggests that while the structure is close to the experimental pose, it is not yet a definitive model. DASAT is a control ligand with a high binder probability. Its predicted potency is 12.41, placing it first among the ligands tested. However, interaction analysis was not run, so the exact nature of its interactions with the receptor is unknown.	Overall confidence is below the well-determined threshold, indicating that the structure may not be as reliable as expected. This caution should be taken when considering the model for further research or practical applications.	model
`ABLAP`	caution	The confidence score for this structure is low. This means it is difficult to trust this structure, and it is likely to be incorrect.	Overall confidence is below the well-determined threshold and the ligand is poorly placed relative to the receptor	model