Coefficient of variation filtering for transitions #35
Comments
|
Can you provide an example where you show all transition plots for single
protein?
…On Thu, Oct 26, 2017 at 6:42 PM Yaamini Venkataraman < ***@***.***> wrote:
Assigned #35 <#35>
to @sr320 <https://github.com/sr320>.
—
You are receiving this because you were assigned.
Reply to this email directly, view it on GitHub
<#35 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AEPHt2JBJ7WzjErfnRqtowY1TbjYz1UXks5swTUIgaJpZM4QIb_c>
.
|
|
For Heat Shock Protein, just sites (WITH TITLES). CV >20 Filtering:
CV >15 Filtering:
CV >10 Filtering:
|
|
ALL FIXED Hang on, just realized that none of my box plots have titles. Give me a few minutes and I can paste the same graphs but with titles that indicate what protein/peptide/transition you're looking at |
|
That would be great
…On Thu, Oct 26, 2017 at 7:00 PM Yaamini Venkataraman < ***@***.***> wrote:
Hang on, just realized that none of my box plots have titles. Give me a
few minutes and I can paste the same graphs but with titles that indicate
what protein/peptide/transition you're looking at
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#35 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AEPHtz_-raoxAVhxXZtKrTsqQdR9VREkks5swTjugaJpZM4QIb_c>
.
|
|
Related to the other question - take the <20 filtered data and determine
how many transitions have data for all samples
On Thu, Oct 26, 2017 at 7:22 PM Steven Roberts <sr320@u.washington.edu>
wrote:
… That would be great
On Thu, Oct 26, 2017 at 7:00 PM Yaamini Venkataraman <
***@***.***> wrote:
> Hang on, just realized that none of my box plots have titles. Give me a
> few minutes and I can paste the same graphs but with titles that indicate
> what protein/peptide/transition you're looking at
>
> —
> You are receiving this because you were mentioned.
>
>
> Reply to this email directly, view it on GitHub
> <#35 (comment)>,
> or mute the thread
> <https://github.com/notifications/unsubscribe-auth/AEPHtz_-raoxAVhxXZtKrTsqQdR9VREkks5swTjugaJpZM4QIb_c>
> .
>
|
|
only 1 |
|
Are you using zeros for values when you have no data? |
|
yes |
|
I'm still concerned about the replication. If the exact same sample is injected in the same mass spectrometer two different times, the results (especially for SRM!) should look the same. With the kind of variance you have, I worry that 1) sample names/locations got mixed up, 2) there is something very wrong with the list of transitions and they are completely unreliable, or 3) the column switch resulted in poor replication. I believe you have ruled out #3 and #1. Right now you are trying to get to the bottom of #2. If it really is a problem with the oyster transitions themselves, then I would expect excellent replication among the PRTC peptides despite poor replication among the oyster peptides. We do not see that. That makes me think there is some kind of technical issue that we have not figured out. |
|
Shoot, the #s should not have been linked to previous issues. That was a mistake! |
|
Emma - lets get together and discuss. Are you available today.
It would appear to me that Laura’s data in not much different than
Yaamini’s.
…On Fri, Oct 27, 2017 at 8:37 AM Yaamini Venkataraman < ***@***.***> wrote:
yes
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#35 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AEPHt9pYTejS1-d5n2toTuxGmtinBWIkks5swfgigaJpZM4QIb_c>
.
|
|
Yaamini - if the “yes” was to my question, that is the wrong way to do it.
It should be treated as no data.
On Fri, Oct 27, 2017 at 8:42 AM Steven Roberts <sr320@u.washington.edu>
wrote:
… Emma - lets get together and discuss. Are you available today.
It would appear to me that Laura’s data in not much different than
Yaamini’s.
On Fri, Oct 27, 2017 at 8:37 AM Yaamini Venkataraman <
***@***.***> wrote:
> yes
>
> —
> You are receiving this because you were mentioned.
>
>
> Reply to this email directly, view it on GitHub
> <#35 (comment)>,
> or mute the thread
> <https://github.com/notifications/unsubscribe-auth/AEPHt9pYTejS1-d5n2toTuxGmtinBWIkks5swfgigaJpZM4QIb_c>
> .
>
|
|
I can join by Hangout assuming my energy stays up. I had surgery on Tuesday but I'm feeling OK today. |
|
@sr320 the yes was to your comment. I had to replace NAs with zeros for my NMDS, so let me look at my dataset with NAs and get back to you? |
|
Thus NMDS is inaccurate also
…On Fri, Oct 27, 2017 at 8:46 AM Yaamini Venkataraman < ***@***.***> wrote:
@sr320 <https://github.com/sr320> the yes was to your comment. I had to
replace NAs with zeros for my NMDS, so let me look at my dataset with NAs
and get back to you?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#35 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AEPHt3froxWt0aCpMIHmCWk9oMEzSIjdks5swfqZgaJpZM4QIb_c>
.
|
|
I actually can't make an NMDS with NA values, so I'm unsure how to correct that? |
|
No action needed- just focus on distinct protein expression visualizations
On Fri, Oct 27, 2017 at 10:37 AM Yaamini Venkataraman < ***@***.***> wrote:
I actually can't make an NMDS with NA values, so I'm unsure how to correct
that?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#35 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AEPHt-hLcPFxGZ45GxA1ZH4WhIGHo1olks5swhScgaJpZM4QIb_c>
.
--
Steven B. Roberts
Kenneth K. Chew Endowed Professor
School of Aquatic and Fishery Sciences
University of Washington, Seattle WA
sr320@uw.edu - 206.866.5141
robertslab.info - @sr320
|
For each pair of technical replicates, I removed any transitions that had a coefficient of variation greater than a certain thresh.old (from Steven's suggestion in #18). I used three different thresholds: CV>20, CV>15, CV>10. I reexamined my technical replication, my NMDS for samples/eelgrass, and my ANOSIMs. I also made box plots for each transition, using just sites and both sites and eelgrass conditions. All of my notebooks can be found here:
CV > 20 filtering results
CV > 15 and CV > 10 filtering results
Boxplots
As I mention in my box plot notebook entry, my biggest concern now is the fact that we selected individual transition data to remove for each technical replicate, as opposed to removing a transition completely from the analysis. So some transitions have a much larger dataset than others. How can I account for these irregularities moving forward? I'm assuming my next step is to also examine the boxplots and quantify how similar/different my median/mean values are (i.e. t-tests or ANOVAs).
Thoughts/suggestions?
The text was updated successfully, but these errors were encountered: