Output invariant sites and nucleotide frequencies

[EpiDemos82]

In general, phylogenetic programs use invariant sites for likelihood calculations. However, a number of programs, such as RAxML and BEAST, can perform ascertainment bias corrections given the number of invariant sites and the frequencies of nucleotides in the alignment. If SNP-sites output these values, they could be used as direct inputs for RAxML, for example.

[tseemann]

I second this suggestion.

Either add a -s (stats?) option to report all sorts of columnar statistics, characters used etc.

OR

Always output this to stderr as part of the logs.

[andrewjpage]

Grand, give me a toy example and I'll sort it out

…

On 27 September 2017 at 08:53, Torsten Seemann ***@***.***> wrote: I second this suggestion. Either add a -s (stats?) option to report all sorts of columnar statistics, characters used etc. OR Always output this to stderr as part of the logs. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#62 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AABeVy1rlA4xOP4bxeVn-EM0LNPbtoOhks5smf7igaJpZM4PhZS0> .

[EpiDemos82]

From your example in the README:

sample1
AGACACAGTCAC
sample2
AGACAC----AC
sample3
AAACGCATTCAN

-s (or stderr) would produce:

Input stats:
Alignment length: 12
Proportion Ns: 0.03
Proportion Gap sites: 0.11
Nucleotide frequencies (A,G,C,T): 45.2,12.9,32.3,9.7

Output stats:
SNP alignment length: 3
Number Gap sites (-) introduced: 1
Proportion gap sites: 0.11

Thinking about this more, I obviously came up with a couple other useful stats. seqtk comp can produce similar stats, but having them in one tool with the speed of SNP-sites would be great.

[tseemann]

But output in a machine readable format so we can parse or JSON-ate.

[slvrshot]

@andrewjpage

Hello. I was curious if this was ever implemented?