ENH: distance decay notebook by Ashkaan #191
Conversation
| "library(dplyr)\n", | ||
| "\n", | ||
| "## I was given a 1-column csv file with sample IDs to include in my analyses\n", | ||
| "use <- read.csv('filepath/file.csv', header = TRUE)\n", |
There was a problem hiding this comment.
Is this the single sample file? If so, can we integrate with the current repo structure?
Fixed a minor typo
| "source": [ | ||
| "The next section performs individual analyses at the state-level, and performs Benjamini-Hochberg correction for multiple comparisons and spits them out into a summary table, so that you can evaluate which states (if any) display significant distance-decay relationships and plot those particular states, if you want.\n", | ||
| "\n", | ||
| "In the first version of the data I recieved, I had to manually clean up multiple discrepancies in state namings (e.g., state labels for Colorado were inputted as both 'CO' and 'Colorado'. Not sure if this has been corrected in the new data. If not, let me know and we can figure out a quick solution." |
There was a problem hiding this comment.
The state names are corrected in the new version, so we shouldn't need to update. But, we're currently storing state information for states outside of the US and that can give weird results.
There was a problem hiding this comment.
That sounds good. Is there anything I can do on my end to help?
AF
On Mon, Jan 25, 2016 at 9:02 AM, J W Debelius notifications@github.com
wrote:
In ipynb/agp_distance_decay.ipynb
#191 (comment):
- " panel.border = element_blank(), panel.grid.minor = element_blank(), \n",
- " panel.background = element_blank()) +\n",
- " theme(axis.line = element_line(color = 'black'))\n",
- "\n",
- "## and we perform a permutational test to assess significance\n",
- "mant.mod <- mantel(comm.dist, gc.dist, method = 'spear', permutations = 999)\n",
- "mant.mod"
- ]
- },
- {
- "cell_type": "markdown",
- "metadata": {},
- "source": [
- "The next section performs individual analyses at the state-level, and performs Benjamini-Hochberg correction for multiple comparisons and spits them out into a summary table, so that you can evaluate which states (if any) display significant distance-decay relationships and plot those particular states, if you want.\n",
- "\n",
- "In the first version of the data I recieved, I had to manually clean up multiple discrepancies in state namings (e.g., state labels for Colorado were inputted as both 'CO' and 'Colorado'. Not sure if this has been corrected in the new data. If not, let me know and we can figure out a quick solution."
The state names are corrected in the new version, so we shouldn't need to
update. But, we're currently storing state information for states outside
of the US and that can give weird results.—
Reply to this email directly or view it on GitHub
https://github.com/biocore/American-Gut/pull/191/files#r50722181.
Ashkaans ipynb
| "## compute Canberra distances\n", | ||
| "comm.dist <- vegdist(t(dat.rel.trans), method = 'canberra')\n", | ||
| "comm.dist <- as.matrix(comm.dist)\n", | ||
| "\n", |
There was a problem hiding this comment.
Based on the notes that came in october, this used to be Bray-Curtis distance. Is there a reason to use Canberra here over Bray-Curtis or UniFrac?
There was a problem hiding this comment.
I thought Canberra had better behaved residuals. Can revert to Bray if you prefer.
No description provided.