There is another example in the CSV-file where already existing stereo-info in the substrate is still lost in the product.
Examples are also provided as short, self-contained Java Unit Test:
https://github.com/mguetlein/TestAmbitSmarts/blob/master/src/test/java/org/kramerlab/test/TestAmbit.java
String smirks = "[#8:1]([H])-[#6:2](-[#6:9](-[#8-:10])=[O:11])=[#6:3](-[#1,#6,#17:12])-[#6:4]=[#6:5]-[#6](-[#8-])=O>>[#8-:10]-[#6:9](=[O:11])-[#6:2](=[O:1])-[#6:3](-[#1,#6,#17:12])-[#6:4]=[#6:5]"; String smi = "C(=C(/C(=O)[O-])\\Cl)/C=C(\\C(=O)[O-])/O"; String expectedSmiles = "[O-]C(=O)C(=O)C\\C=C\\Cl";
hmm, cannot change the title: "Stereochemistry of substrate lost in product"
envipath ids:
rule3138
u138720
Already fixed in Ambit-Smarts library
Fixed in 3.0.3-SNAPSHOT latest revision 7696