inconsistency in 1-mer substitution matrices
HH_S1F, obtained by normalizing HS1FDistance (which is symmetric) by rows, is symmetric.
However, HKL_S1F and MK_RS1NF, produced by createSubstitutionMatrix(...returnModel="1mer"...), and reported in Cui et al., 2016, are not symmetric.
Need to find out why and update HH_S1F accordingly.
Previous documentation for HKL_S1F and MK_RS1NF saying that they are symmetric have been fixed.
Comments (7)
-
-
And we should update
shazam::calcTargetingDistance
accordingly as well. -
- Rebuild
HH_S1F
from Yaari, 2013 S5F (or Namita's email). - Delete
M1N
. Update docs ofMK_RS1NF
to indicate it replacesM1N
.
- Rebuild
-
From @guryaari:
# HS1F (all patients combined): A C G T A 0 25080 39261 17697 C 21319 0 26449 39500 G 81060 51891 0 26768 T 17620 39261 22516 0 # Normalized: A C G T A 0.0000000 0.2442934 0.5075163 0.2219227 C 0.3057120 0.0000000 0.3248893 0.4944897 G 0.4785709 0.3030779 0.0000000 0.2835875 T 0.2157171 0.4526287 0.1675943 0.0000000 # Symmetrized: A C G T A 0.0000000 0.2343033 0.4874240 0.2782727 C 0.2343033 0.0000000 0.2782727 0.4874240 G 0.4874240 0.2782727 0.0000000 0.2343033 T 0.2782727 0.4874240 0.2343033 0.0000000 # 1/Symmetrized (for distance): A C G T A 0.000000 2.080312 1.000000 1.751606 C 2.080312 0.000000 1.751606 1.000000 G 1.000000 1.751606 0.000000 2.080312 T 1.751606 1.000000 2.080312 0.000000
-
reporter @javh For HS1FDistance and HH_S1F, is the goal to delete and replace HS1FDistance with the updated version of HH_S1F (just like we're deleting and replacing M1NDistance with MK_RS1NF)?
-
@jqz Yes.
-
reporter - changed status to resolved
This commit: 1) updated HH_S1F [now non-symmetric]; 2) deleted HS1FDistance, replaced with updated HH_S1F, made a note of replacement in HH_S1F's doc; 3) deleted M1NDistance, replaced with MK_RS1NF, and made a note of replacement in MK_RS1NF's doc
- Log in to comment
Also, distances are calculated from targeting values for 5-mer models and from substitution values for 1-mer models.