Dear HADDOCK support team,
Could you please inspect the backend state of these three small protein-peptide refinement jobs? All three remain Running, with it0 / it1 / itw = 0.0% / 0.0% / 0.0%, more than 26 hours after submission. They already showed the same state during a previous check on 21 October. There is no visible error and no CNS/topology log exposed on the running pages.
| Job | Submitted / started as displayed in Workspace | Link |
|---|---|---|
| 784866 (ACP09_1IVO_M1_HA12_20261021) | 2026-10-21 02:49:00 | run page |
| 784870 (ACP04_1YY9_M3_HA12_20261021) | 2026-10-21 02:54:00 | run page |
| 784873 (ACP05_1YY9_M3_HA12_20261021) | 2026-10-21 03:07:00 | run page |
Workspace does not label the timezone of the displayed timestamps above. Last check: 2026-10-22 05:49 UTC.
Each job has a single pre-positioned receptor-peptide pair, with 12 / 12 / 12 models requested for the three stages. Receptors are EGFR ectodomain fragments (511 residues for 1IVO, 613 for 1YY9); peptides have 12 or 13 standard amino acids. The native receptor disulfides should be retained. We submitted through Submit file, preserving the original parameters (no protocol preset override). The downloaded server-accepted JSON matched every submitted field, including the input coordinates.
Relevant settings common to these pilots:
delenph = false
structures_0 = 12; structures_1 = 12; waterrefine = 12
randorien = false; rigidmini = false; rigidtrans = false
crossdock = false; rotate180_0 = false; rotate180_1 = false
fix_origin = true for both partners
top_file = protein-allhdg5-4.top
par_file = protein-allhdg5-4.param
receptor link_file = protein-allhdg5-4-noter.link
peptide link_file = protein-allhdg5-4.link
peptide charged_nter = true; charged_cter = true
tadhigh_t = 2000
initiosteps/cool1_steps/cool2_steps/cool3_steps = 2000/2000/4000/4000
timestep = 0.002; tadfactor = 8
solvshell = false
ssdihed = none; dihedrals_on = false
The change to delenph=false was deliberate: in an earlier completed run we found zero HA atoms in the actual complex PSF, whereas the supplied topology defines the C-alpha chirality improper using HA/N/C/CB (and Thr/Ile C-beta impropers use HB). Some earlier structures developed stereochemistry defects. We repaired affected input side chains against the original crystal structures, preserved the peptide pose and backbone, and screened the serialized inputs for alpha/beta stereochemistry, severe heavy-atom clashes, peptide bonds and native disulfides. One pilot starts from an input already normal in alpha/beta stereochemistry. This is a hypothesis-testing pilot; we have not established that retaining hydrogens fixes the defects, and we do not know whether that setting is related to the current stall.
Could you please check:
- Whether these jobs actually reached a worker: scheduler/grid dispatch IDs, pending reason, worker start/last activity, retrieval state, and any stale per-user concurrency slot.
- Whether topology generation completed:
begin/*.out, PSF generation, hydrogen/terminal patches, and any unsupported all-hydrogen topology or parameter errors. - The driver log and the first CNS
it0_refine_*.out(or compressed equivalent), especially missing parameters, atom limits,ERRmessages, crash/retry loops, or failed result retrieval. Is this 12-model setup valid on the current server? - Whether this is infrastructure-related or input/protocol-related, and the exact corrective action needed. Please retain the existing jobs and original diagnostic/results files; let us know before cancelling or deleting them. If an in-place resume is possible, that is preferable.
We will avoid duplicate submissions or bulk expansion while this remains unresolved. If a corrected rerun is needed, we will first run only one small pilot and validate the actual topology, termini, chirality, clashes, peptide bonds and disulfides before expanding. A redacted error excerpt or the relevant diagnostic logs would be very helpful.
Thank you.