@agakrawczyk and me ran in problems when using bowtie2 and slimm:
Yara, bowtie2 and slimm indizes with Human chr1 & chr11 and C-RVDB (https://rvdb.dbi.udel.edu/) were built.
An in silico data set comprising 91% Human chr1 & chr11 reads and 9% Human virus reads of various species was generated and mapped with bowtie2 or yara.
Slimm was used for abundance estimation for bowtie2 or yara mappings.
While "yara+slimm" gave consistent results for all assayed viruses, "bowtie2+slimm" did fail for two viruses. Closer inspection on mapping files showed that bowtie2 reported many discordant mapping across various reference sequences:
Dear @temehi & @agakrawczyk ,
@agakrawczyk and me ran in problems when using bowtie2 and slimm:
While "yara+slimm" gave consistent results for all assayed viruses, "bowtie2+slimm" did fail for two viruses. Closer inspection on mapping files showed that bowtie2 reported many discordant mapping across various reference sequences:
Yara did not report these discordant mappings.
My question: How does slimm deal with discordant mappings? My suspection is that these are discarded.
Thanks in advance!