PolymerLit

The data repo for polymer image recognition

PolymerLit-MT

This is the subset containing 300 images with corresponding molblocks from the BigSMILES Machine Translation (MT) paper by Deagen et al [1]. The raw data have been made available publicly by the authors. There were some minor inconsistencies in the molblocks where aromatic bonds were recorded as type 4 (i.e., non-kekulized) but the images actually displayed alternating single and double bonds (i.e., kekulized). We have manually corrected all these blocks.

PolymerLit-Olsen

This is the subset containing 468 images with corresponding molblocks from 3 publications by the Olsen group at MIT [2,3,4]. These manuscripts and SIs are all open-access, but the images were originally unlabeled. We used our PolymerScribe model to predict the molblocks which were manually corrected afterward.

PolymerLit-OA

This is the subset containing 1,000 images with corresponding molblocks from Open-Access (OA) articles. We have carefully chosen the images to be from only articles with CC-BY-NC-ND and less restrictive licenses. Because of the ND clause (Non-Derivative) for some, we decided to release the images exactly as how they appear originally, together with the coordinates of the bounding polygons surrounding the polymer structures. These bounding polygons were also drawn manually with the help of the open-source tool CVAT.ai, and the "cropped" images can be easily reconstructed with the provided script,

$ pip install pillow
$ python generate_cropped_images.py

The "cropped" images will be populated under PolymerLit-OA_processed, whose filenames should match the provided molblocks (doi_suffix.corrected.mol) which were predicted by PolymerScribe and manually corrected in a similar manner. The references for all images and their licenses are recorded in PolymerLit-OA_refs.xlsx.

Canonical BigSMILES

canonical_bigsmiles.tsv holds the BigSMILES ground truth for every *.corrected.mol in this repo, one row per molblock:

column	meaning
path	the *.corrected.mol path, relative to the repo root
bigsmiles	BigSMILES converted from the molblock
canonical_bigsmiles	its canonical form
status	how the canonicalization went (below)

status	meaning	count
SUCCESS	canonicalized, and the string changed	1524
NOOP	already canonical	9
FAIL	canonicalization failed	188
BIGSMILES_FAILED	no BigSMILES could be obtained at all	47

Only SUCCESS and NOOP carry a usable canonical form (1533 of 1768). The status column exists because a failed canonicalization returns its input unchanged, exactly as an already-canonical molecule does -- so canonical_bigsmiles == bigsmiles alone cannot tell the two apart.

Atoms that are placeholders rather than real chemistry -- an atom carrying an A alias record (an abbreviation such as Ph, CF3 or TEG), or one drawn as a bare R -- are normalized to a single wildcard element, [Y], before conversion. Without this the same structure converts differently depending on whether it was drawn in ChemDraw (carbon + alias) or produced by a model (element R), and [R] cannot be parsed as an atom at all, so it never canonicalizes. * atoms are left alone: here they are polymer attachment points and become bonding descriptors.