Spaces:
Sleeping
Sleeping
Commit History
8 days ago
-
Refresh the paper list
29dd4f0
15 days ago
-
Refresh the paper list
843f017
22 days ago
-
Refresh the paper list
f056728
29 days ago
-
Refresh the paper list
0bbc453
-
Refresh the paper list
c5b9c94 -
Refresh the paper list
3eb6429 -
Refresh the paper list
368d0ab
-
Read the catalog's screening decisions; refresh the list against today's catalog
harvest.py takes --decisions screening_decisions.tsv, the catalog's record of papers its bulk screenings rejected or held, and drops them like the watch list's DOIs. The list is refreshed against the current catalog: the six papers added from the last harvest are now catalogued, the pi-HelixNovo erratum is dropped as rejected, and the header no longer reads "since 20". Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011PuBZJGkLScPh6zuMyYXh74c38c86
-
Refresh the paper list
fe38264
-
Close the loop: carry papers over, drop the ones rejected
The Space was re-proposing work already argued about, because a rejected paper has no catalog DOI and so came back as new every week. harvest.py now reads the catalog's WATCHLIST.md and drops those by DOI, the same contract build_candidates.py has had. Matching is DOI-only on purpose: a watch-list entry argues about a paper in a sentence, and fuzzy-matching sentences against titles would drop papers nobody rejected. The other half already worked. A paper added to the catalog is matched by DOI or normalised title and shows as catalogued, so it stops being new by itself. Fixing the first half exposed a worse bug. harvest.py REPLACED the list, and two runs a week apart fetched 682 and 435 records over the same window, because neither API returns a stable set. An overwrite would have dropped twenty papers from the Space for no reason anyone could see. The list is now cumulative: previous entries are carried forward and re-filtered, so it shrinks only when a paper is catalogued or rejecte…e5af595 -
Refresh the paper list
7f67948 -
Refresh the paper list
5e76a58 -
De Novo Peptide Sequencing Radar
Recent de novo peptide sequencing papers, cross-checked against the Awesome De Novo Peptide Sequencing catalog so that what is missing from it is visible. harvest.py queries Europe PMC and OpenAlex and then does the part that matters: separating de novo SEQUENCING of peptides from de novo DESIGN, which is a different field sharing a Latin phrase and which dominates any keyword feed, and from de novo sequencing of other analytes (glycans, oligonucleotides, RNA, DNA, transcriptomes). Data and model deposits are dropped as well: Zenodo weights, SciLifeLab checkpoints and MassIVE records match the topic perfectly and are not papers. On the first run: 682 records fetched, 616 after deduplication, 551 dropped as not de novo peptide sequencing, 14 as deposits, 5 as another analyte, 1 as design, leaving 45 of which 35 were already catalogued and 10 were new. app.py only displays data/papers.json. The harvest runs in a scheduled GitHub Action that commits here, so the data is versioned rathe…2136560