NCATS (chemical) ocr engine that can vectorize chemical images into Chemical objects preserving the 2D layout as much as possible. The code is still very raw in terms of utility. Check out https://molvec.ncats.io for a demonstration of MolVec.
The easiest way to start using Molvec is to include it as a dependency in your build tool of choice. For Maven:
<dependency>
<groupId>gov.nih.ncats</groupId>
<artifactId>molvec</artifactId>
<version>0.9.8</version>
</dependency>
File image = ...
String mol = Molvec.ocr(image);
MolVec supports asynchronous calls
CompleteableFuture<String> future = Molvec.ocrAsync( image);
String mol = future.get(5, TimeUnit.SECONDS);
Since version 0.9.8, Molvec has a more robust API that allows for creating both molfiles and SDfiles and adding properties to the SDfiles.
MolvecOptions options = new MolvecOptions()
.setName(name)
.center(true)
.averageBondLength(2);
MolvecResult result = Molvec.ocr(f, options);
//write out a SDfile without any properties
String sdfile = result.getSDfile().get());
//create a map of key-value pairs and include as properties in an SDfile
Map<String, String> props = new HashMap<>();
props.put("File Name", f.getName());
String sdfileWithProperties = result.getSDfile(props).get();
The Molvec jar has a runnable Main class with the following options
usage: molvec ([-gui],[[(-f <path> [-o <path>]) | (-dir <path> [[-outDir <path> | -outSdf <path>]],[-parallel <count>])]],[-scale
<value>],[-h])
Image to Chemical Structure Extractor Analyzes the given image and tries to find the chemical structure drawn and
convert it into a Mol format.
options:
-dir <path> path to a directory of image files to process. Supported formats include png, jpeg, tiff and
svg. Each image file found will be attempted to be processed. If -out or -outDir is not
specified then each processed mol will be put in the same directory and named
$filename.molThis option or -f is required if not using -gui
-f,--file <path> path of image file to process. Supported formats include png, jpeg, tiff. This option or -dir
is required if not using -gui
-gui Run Molvec in GUI mode. file and scale option may be set to preload file
-h,--help print helptext
-o,--out <path> path of output processed mol. Only valid when not using gui mode. If not specified output is
sent to STDOUT
-outDir <path> path to output directory to put processed mol files. If this path does not exist it will e
created
-outSdf <path> Write output to a single sdf formatted file instead of individual mol files
-parallel <count> Number of images to process simultaneously, if not specified defaults to 1
-scale <value> scale of image to show in viewer (only valid if gui mode AND file are specified)
Examples:
$molvec -f /path/to/image.file
parse the given image file and print out the structure mol to STDOUT
$molvec -dir /path/to/directory
serially parse all the image files inside the given directory and write out a new mol file for each image named
$image.file.mol the new files will be put in the input directory
$molvec -dir /path/to/directory -outDir /path/to/outputDir
serially parse all the image files inside the given directory and write out a new mol file for each image named
$image.file.mol the new files will be put in the directory specified by outDir
$molvec -dir /path/to/directory -outSdf /path/to/output.sdf
serially parse all the image files inside the given directory and write out a new sdf file to the given path that
contains all the structures from the input image directory
$molvec -dir /path/to/directory -outSdf /path/to/output.sdf -parallel 4
parse in 4 concurrent parallel thread all the image files inside the given directory and write out a new sdf file to
the given path that contains all the structures from the input image directory
$molvec -dir /path/to/directory -parallel 4
parse in 4 concurrent parallel threads all the image files inside the given directory and write out a new mol file for
each image named $image.file.mol the new files will be put in the directory specified by outDir
$molvec -gui
open the Molvec Graphical User interface without any image preloaded
$molvec -gui -f /path/to/image.file
open the Molvec Graphical User interface with the given image file preloaded
$molvec -gui -f /path/to/image.file -scale 2.0
open the Molvec Graphical User interface with the given image file preloaded zoomed in/out to the given scale
Developed by NIH/NCATS
Molvec Comes with a Swing Viewer you can use to step through each step of the structure recognition process
You can report issues or feature requests either by creating issue tickets on our github page or by forwarding questions and/or problems to [email protected]