In reference to the below comment about the new release of Alevin-fry, we should incorporate being able to use the -unfiltered parameter. This should eliminate identifying any barcodes outside the possible list of barcodes in the 10X whitelist.
Additionally, we will need to update the docker container to use the new release of Alevin-fry.
Hi all,
I'll just note that we've recently released v0.2.0 of alevin-fry, and it's been propagated through bioconda as well. In addition to general improvements to the runtime profile (vastly improved memory usage at very high thread count), this release also adds initial (beta) support for quantification via an "unfiltered" permit list using the -u option to the generate-permit-list command; this functionality was previously unsupported. This flag will allow one to properly pass e.g. the 10x permit list to the generate-permit-list so that one can then quantify all cells, rather than just those selected by the knee threshold. There is also --min-reads command that determines the minimum number of reads (not UMIs, as this is pre-quantification) a cell barcode needs to have to be considered present. The default is 10, but you can set it at 1 if you really want to quantify everything. Let me know if you need any help or support running alevin-fry on the testing data, or if you have any questions about the different resolution options (e.g. https://github.com/AlexsLemonade/alsf-scpca/pull/60#discussion_r611078240).
In reference to the below comment about the new release of Alevin-fry, we should incorporate being able to use the -unfiltered parameter. This should eliminate identifying any barcodes outside the possible list of barcodes in the 10X whitelist.
Additionally, we will need to update the docker container to use the new release of Alevin-fry.
Hi all,
I'll just note that we've recently released v0.2.0 of
alevin-fry, and it's been propagated through bioconda as well. In addition to general improvements to the runtime profile (vastly improved memory usage at very high thread count), this release also adds initial (beta) support for quantification via an "unfiltered" permit list using the-uoption to thegenerate-permit-listcommand; this functionality was previously unsupported. This flag will allow one to properly pass e.g. the 10x permit list to thegenerate-permit-listso that one can then quantify all cells, rather than just those selected by the knee threshold. There is also--min-readscommand that determines the minimum number of reads (not UMIs, as this is pre-quantification) a cell barcode needs to have to be considered present. The default is 10, but you can set it at 1 if you really want to quantify everything. Let me know if you need any help or support running alevin-fry on the testing data, or if you have any questions about the different resolution options (e.g. https://github.com/AlexsLemonade/alsf-scpca/pull/60#discussion_r611078240).Best, Rob
Originally posted by @rob-p in https://github.com/AlexsLemonade/alsf-scpca/issues/63#issuecomment-824228309