motifmatchr fails when using some motifs collections (e.g. https://doi.org/10.1093/nar/gkad1240 ) due to small rounding errors in the column sums of the PFM preventing pwmType from determining whether the motifs are PFM or PWM. While it would be possible to process the matrices to avoid this, it's actually not that trivial and unnecessary, because the problem is easily solved by increasing the tolerance in pwmType.
There's also no drawback in doing so, given that it's virtually impossible that all the colSums of all PWM are by chance within 0.99-1.01.
motifmatchr fails when using some motifs collections (e.g. https://doi.org/10.1093/nar/gkad1240 ) due to small rounding errors in the column sums of the PFM preventing
pwmType
from determining whether the motifs are PFM or PWM. While it would be possible to process the matrices to avoid this, it's actually not that trivial and unnecessary, because the problem is easily solved by increasing the tolerance inpwmType
. There's also no drawback in doing so, given that it's virtually impossible that all the colSums of all PWM are by chance within 0.99-1.01.