These are existing polishing programs and pipelines to compare with Windel. Because each program has its own idiosyncrasies, as well as different outputs, it is necessary to have a document to collate these, so that we can wrangle their respective outputs into a format that we can compare in an unbiased way. This is that document.

https://github.com/broadinstitute/pilon/wiki

Pilon conveniently provides an option to output a changes file, much like Windel's generatewindelchanges.py. This option is --changes for when running Pilon.

https://github.com/isovic/racon

As far as I can tell, Racon does not provide an equivalent to Pilon's --changes option. It will probably require some post-processing of the output FASTA sequences to infer edit coordinates. This may be grounds to exclude it from analysis.

https://github.com/bcgsc/ntEdit

This uses a K-mer based approach, instead of using an alignment. Their paper concludes that it is not as accurate as polishing programs which relied on paired-end mapped reads, so it may not be worth including this in the comparison.

ntEdit does output a changes.tsv file which should help with comparisons.

https://github.com/Nextomics/NextPolish

NextPolish has a -debug option, which outputs details of polished bases to stderr. I have not tested this yet, but I hope that this contains some equivalent of Pilon's --changes output.

https://github.com/EichlerLab/indel_correction_pipeline

This is unpublished, but probably worth including, as I think using FreeBayes to address this problem is relatively common.

Since it uses the output of FreeBayes (a vcf file) as input, it shouldn't be too hard to generate a changes file for comparison with the other programs.