No output for some species about fur HOT 17 CLOSED

ok, i follow- i am going to use phylonium to validate my other target/neighbourhood designations, it is clearly far more suited to the task at hand than just grepping through taxids in the NCBI genbank taxids.

i will leave this open for the time being but hopefully once i have completed the above i will be able to close this and the other issue.

i would like to thank you again for your patience and for taking the time to address my problems. i recognise these were due to a lack of knowledge on my behalf (i'm pretty new to taxonomy/phylogeny!!) and not due to shortcomings of the software.

from fur.

Thanks for going to this length comparing fur and classical alignments. Glad to see it all makes sense. I'm more than happy to leave it at that.

from fur.

from fur.

i can go (hopefully) one better and provide the links i used to download them. these are for strep pyogenes

ftpfilepaths_genbank.txt
ftpfilepaths_genbank_neighbourhood.txt

from fur.

and this is the target for Rickettsia prowazekii

ftpfilepaths_genbank_target_RP.txt

my neighbourhood folder was empty because there were no close relative genomes fitting my criteria, i assume this doesn't affect the output of the first step, but if I can get fur running so that it outputs things from the last step, I'll obviously need a few neighbourhood files.

let me know if you need anything else, i'd be really happy to get this running!

from fur.

from fur.

thank you! i will check if this is for the case for my others.... it shouldn't be because i filtered by species/close relative tax ids. i'll let you know if the problem persists after i've checked for duplication, it did not occur to me this could have happened so i feel a bit silly now.

any idea what the issue could be for rickettsia?

from fur.

from fur.

Let me do some checking- i need to find out what caused the duplication in my source/target dirs. i automated for every species, manually rebuilding the rickettsia db and running does indeed get half a megabase of sequence. if this is all down to bugs in my automation script then i sincerely apologise for wasting your time!!

it will take me a while to check things manually but i hope to get back to you by the end of today (UK time). thank you for checking this for me.

from fur.

from fur.

from fur.

brilliant- thank you!

i have been going through everything manually, and in some cases i had duplicates between target and neighbourhood and am now getting results.

in others i can't see any problems. I have added an example from Clostridium botulinum plus a screenshot showing how i make the db and run fur. as i've done this manually i unfortunately don't have the files with all the ftp links. I get similar for Chlamydia pneumoniae.

neighbourhood.txt
target.txt

from fur.

I took a look at your Chlamydia tree using
phylonium neighbors/*.fna targets/*.fna | clustDist | midRoot | new2view
from this website and got - targets in black, the neighbor in red

If I move the outlying targets into the neighborhood, where they seem to belong, I get 23 kb output.

from fur.

fur should be as self-explanatory as possible, and your comments pushed it a bit further in that direction. So I thank you for taking the time to get in touch. If you find anything else while going through your analyses, please let me know.

from fur.

I will, i'm glad its been useful. I am getting on much better by validating my initial species lists using the phylonium command you mentioned above.

Do you have a recommendation for species that are very close to their close relatives?

For example phylonium gives the attached matrix for the first chromosome of brucella suis vs canis, and the distances between the two species are significantly smaller than for other species I've tested. The visualisation shows that they have been assigned to different branches, so they are separable. But very little genetic material is found to be unique in suis (like < 1 kb), do we just conclude that this is all that distinguishes between them? From alignments between sequences from the two species I would expect more than I get from fur, and can do if I reduce the window size significantly but am uneasy about this.

I will be consulting my lab about this as well, but if you have any input from the perspective of fur I would be grateful to hear it.

phylonium_output.txt

from fur.

That's an nice example, Brucella suis in black, B. canis in red:

And as you said, there is little output. B. suis as target gives 1433 bp with default values and B. canis 618 bp, which increases to 707 bp if megablast is used instead of standard blast (-m). Even a window size of 10 didn't yield much more in my hands - it just slowed down the search. You might try running makeFurDb with other target representatives (by default it's the longest target), but my suspicion is, that also won't change much. Do you have an example for a region found by alignment and missed by fur?

from fur.

Apologies for the delay in my reply- I have been going through with a fine toothed comb and it appears the alignment was a glitch in my viewer as when using a different one it is not present.

I went through the entirety of the alignments to the second chromosome (where there are slightly more differences) and am beyond impressed that phylonium can separate canis from suis, because even by eye the differences are difficult to discriminate.

I think in this case there is genuinely this little to distinguish between the two species and doing anything to increase it would be artificial rather than reflecting a true, biologically meaningful, difference.

With that, I am happy to close this issue unless you would like to say anything else. I would like to thank you again for your help, as a result of your input I will be using fur and related software and recommending it to colleagues and collaborators.

from fur.