Difference between revisions of "HR Clinic"

From Jimenez Group Wiki
Jump to: navigation, search
(→‎Some topics for discussion)
(→‎Some topics for discussion)
Line 157: Line 157:
 
** Donna: small variations in peak tails for Galway data. What is real vs other peaks, and how to deal with it?
 
** Donna: small variations in peak tails for Galway data. What is real vs other peaks, and how to deal with it?
 
** Dara Salcedo: variations of peak shape with m/z across 10-300, and how to deal with it with current software
 
** Dara Salcedo: variations of peak shape with m/z across 10-300, and how to deal with it with current software
−
** Examples of handling 'camels' and dromedaries (?)
+
** Examples of handling 'camels' and dromedaries - v-mode ions appearing in w-mode spectra
  
  
 
* HR-AMS Spectra database
 
* HR-AMS Spectra database
−
** Encourage users to give to Michael directly during meeting. Need peak shape function, sigma parameters, m/z cal.
+
** Encourage users to give to Michael directly during meeting. Need peak shape function, sigma parameters, m/z cal. Donna's post-clinic note: Pete DeCarlo has volunteered to be the coordinator
−
 
+
for the [http://cires.colorado.edu/jimenez-group/HRAMSsd/ HR-AMS Spectra database]
  
 
* Squirrel Topics
 
* Squirrel Topics

Revision as of 10:36, 11 June 2009

Tentative Schedule

The format of the clinic will be for individuals to work on their own datasets, with intermittent group discussion of important topics, and with the experts going around the room and having smaller discussions with individual users or small groups. The detailed order of the topics will follow the listing below in the section Some Topics for Discussion


Sat June 6 (Ekeley E1B20)

  • Themes of the day: Getting started, UMR/Squirrel issues, setting up HR work
  • 9:20 am: Ingrid will meet attendees at Boulder Outlook Hotel lobby and walk to meeting room (they will leave by 9:30 o'clock)
  • 10:00 am: arrival at meeting room, setting up of laptop power and internet (coffee, tea available)
  • 10:15 am: start of meeting - discussion of topics and format
  • 10:45 am: Donna - Distribution and upgrade to the new version of Squirrel and Pika
  • 12:30 pm: lunch (catered)
  • 1:45 pm: continuation of UMR Squirrel issues discussion
  • 3:00 pm: break w/food
  • 3:30 pm: setting up HR fits: m/z calibrations

Sun June 7 (Ekeley E1B20)

  • Theme of the day: Setting up HR work II
  • 9:30 am: discussion of peak width and peak shape in PIKA (coffee, tea available)
  • 10:30 am: break w/food
  • 12:30 - 1:30 pm: lunch on your own
  • 3:30 pm: break w/food
  • 6:30 pm: head back to hotel before dinner if you want
  • 7 pm: Head downtown from Ekeley for Group dinner at The Med (hosted by ARI)
  • The Med Mediterranean Family Style Restaurant, reservation at 7:30pm

Directions to Dinner:

Mon June 8 (CIRES Auditorium -- backup location Hale 260)

  • Theme of the day: HR work w PIKA
  • 9 am: start (coffee, tea available)
  • 10:30 am: break w/food
  • 3:30 pm: break w/food

Tue June 9 (CIRES Auditorium)

Wed June 10 (CIRES Auditorium)


  • 5 pm: end of meeting

Some topics for discussion

Donna's post-clinic notes: We discussed the topics below throughout the clinic, not necessarily in the order presented here. Users are strongly advised to go the the HR-ToF AMS analysis wiki for up-to-date discussion and best practices.


  • General comments about the differences between unit resolution vs high resolution analysis
  • Getting ready for HR fits
    • Baselines
      • New, better option in ver 1.05 is to remove 'squirrel' baselines from raw spectra
        • Removing 'squirrel' baselines ensures consistency of the application of baselines for all runs
        • Pika baselines are dependent upon the choice of HR fragments to fit, are constant, not linearly interpolated.
    • m/z parameters
      • Review the different diagnostics from the m/z panels, go through an example of the math.
      • How to evaluate when a set of m/z parameters is good enough
      • Discuss the newish idea that we should not use elemental values (N,C,O) for m/z fit
      • Discuss why/when open and closed have different parameters
      • Mike Cubison to demonstrate m/z sensitivity tool
    • Peak Shapes
      • Using the squirrel baseline instead of constant Pika-calculated baselines.
      • What the effect of having different left and right-side constant baselines (like for the air peaks)
      • Left- and right-side peak shape options from new panel
      • Discussion of the idea that the peak 'center' (as defined by the gaussian) may look different from the highest point because the peak may look more poisson-like.
      • Discuss the option to hold the peak center and/or peak width when finding peak shape.
      • Modifying the peak shape 'by hand'
      • Discussion of v/w mode peak width and peak shape results.
    • Fitting Masses
      • The selection of masses to fit
      • How to update the 'master' list of all possible species while retaining current mass selections
      • The 'name' of the element, vs it's likely chemical structure, i.e, we use CH3SO2 not CH2SO2H. We need to keep atoms together for APES code to correctly parse
      • Info about current default masses to fit (Users need more guidance to understand what is fit than beyond having to look through all 200+ masses, although it is ideal that users are aware of every peak that they fit).
      • Talk about changes to the default - now fewer masses are fit by default, and users need to add masses if they appear.
      • How users can add HR spectra to the master list of all available HR fragments (and then to be shared with all).
      • Issues with doing HR fits past > 100 m/z.
      • Discuss version change whereby the fit isn't a 'gaussians multiplied by scalars' , rather the more intuitive 'Normalized' Peak Shape is fit via a look-up table.
      • The importance of fitting all peaks that appear, even if they don't appear in the raw difference spectra
      • Adding new families and the difference between the family concept and the HR batch/frag concept
    • Constrain Isotopes
      • When to not constrain, what does it mean when the sticks don't seem to perfectly match the isotopic ratio
      • Issues when dealing with trying to fit masses that are near x.5 m/z (Dara's question)
      • How constrained factor is generated (i.e. isotopic factor of a fragment with more than one C and one 13C atom)


  • Evaluating Pika sticks for One 'Run'
    • Discuss the HR_PeakHeights_gr controls
    • Discuss the issue of the artifact of the widening the averaged raw spectra
    • The diagnostic tools Steps 4A, 4B, 4C, 4D
    • Discuss how Pika/Igor can return a peak height that is < 0 (Igor 'feature')
    • General problems and issues when fitting the major air peaks.
      • How far from main air peak (in the m/z dimension) can we reliably fit?
      • What to do/how to think if there is saturation
    • How to evaluate when to add/remove masses
      • Looking at residuals to determine how to evaluate when to add/remove masses
      • Looking at related peaks (e.g. CxHy vs Cx+1Hy+2)
      • Discussion of how the HR stick Diff matrix saves sticks in columns, not m/z, so adding or removing one mass to fit means redoing everything.
    • Explain the difference between fitting the difference spectra and calculating the difference HR sticks via Open HR sticks - Closed HR sticks.
    • Explain why it is difficult to fit only one set of peaks centered at one integer; instead the default is to fit all integer masses because of isotope dependencies.
    • Delphine: caveats of HR sticks/fits for near neighbour peaks (CxHyNz and CxHyNzOw families)


  • HR Fit 'Errors'
    • Open - Closed Stick 'Counting' Errors (Pete's Errors)
    • Data - Fit 'Residual' Errors (Doug's Errors)
    • How to use mass accuracy to estimate uncertainty
    • Other sources of uncertainties


  • Evaluating Pika sticks for a ToDo wave
    • How to evaluate when to add/remove masses
      • The use of the many parameter panel to identify when a mass needs to be added.
      • Looking at the time series of all chosen fitted masses for one integer m/z to see if there the masses are too close and the fits 'compensate' each other.
    • Doug's 5 panel graph
    • Discuss future incorporation of HR frag table, RIEs
    • Mike Cubison's SQ vs Pika code (for time series)


  • APES
      • How to access/dump the HR sticks.
    • Basis of elemental analysis (was very confusing in Users Meeting)
    • How to deal with CO2, CO, H2O et
    • Evaluation of the quality of N/C, S/C
    • Problems and doubts from users on how to use the software


  • HR PMF
    • New features: seeds, bootstrapping
    • Examples of applications from users


  • PIKA in Real Life (also discussed throughout and within other topics)
    • Users bring up doubts, examples of problems or unusual data and we discuss as a group
    • Donna: small variations in peak tails for Galway data. What is real vs other peaks, and how to deal with it?
    • Dara Salcedo: variations of peak shape with m/z across 10-300, and how to deal with it with current software
    • Examples of handling 'camels' and dromedaries - v-mode ions appearing in w-mode spectra


  • HR-AMS Spectra database
    • Encourage users to give to Michael directly during meeting. Need peak shape function, sigma parameters, m/z cal. Donna's post-clinic note: Pete DeCarlo has volunteered to be the coordinator

for the HR-AMS Spectra database

  • Squirrel Topics
    • Discussion of ICARRT formatting tool
    • Discussion of QuickView tool

Logistics

  • Visiting us - Link to transportation (from Airport and within Boulder)


    • On Mon-Wed we will meet on the CIRES Auditorium (backup location is Hale 260)
    • Campus Map
    • Jimenez Group Lab is in Ekeley E150
  • Laptop Power and Internet
    • We will provide plug strips for laptops.
    • Please remember to bring plug adaptors if coming from abroad
    • Wireless internet Access will be available for 40 computers, June 3 thru June 12
      • Username: confams
      • Password: Will be given out at the conference
  • Hotel
    • We have a pricing agreement with the Boulder Outlook Hotel link
    • Hotel to Campus Map
    • You can reserve a room with one king bed or two twin beds for $98.00 per night by saying that you are with the HR-AMS Clinic
    • phone numbers: 303.443.3322/800.542.0304
    • Summer in Boulder is a busy time and the hotel does have the possibility of filling up. These rooms are not reserved and the sooner that a reservation can be made the better.
  • Food
    • Lunch and Coffe/Tea Places: Google Map link

Attendees

  • Confirmed Full Time Attendees (40):
    • Doug Worsnop, Manjula Canagaratna, and Sally Ng (ARI)
    • Donna Sueper (ARI & Colorado)
    • James Allan, Rami Alfarra, and Niall Robinson (Manchester) (R)
    • Qi Zhang (SUNY-Albany)
    • Pete DeCarlo (PSI, Switzerland)
    • Mike Cubison, Sanna Saarikoski, Ken Docherty, Ingrid Ulbrich, Carly Robinson, Amber Ortega, Delphine Farmer, & Jose-Luis Jimenez (Colorado)
    • Steve Howell & Lindsey Shank (U. Hawaii)
    • Taehyoung Lee & Laurie Mack (Colorado State)
    • Puneet Chhabra & Jill Craven (Caltech) (R)
    • Lea Hildebrandt (Carnegie Mellon)
    • Samara Carbone and Risto Hillamo (FMI)
    • Qi Chen (Harvard)
    • Toshihide Hikida (Sanyu, Japan)
    • Yongjie Li (Hong Kong and Harvard)
    • Christian (Juelich, Germany)
    • Jurgita Ovadnevaite (NIU Galway, Ireland)
    • Chiara Di Marco & Gavin Phillips (CEH, UK)
    • John Shilling (PNNL)
    • Akinori Takami (NIES, Japan) (R)
    • Dara Salcedo (U Morelos, Mexico)
    • Julia Schmale (MPI Mainz)
    • Manuel Dall'Osto (Galway)
    • Liz Alexander (PNL)
    • Ping Cheng (Handix)
  • Part Time Attendees (2)
    • Roya Bahreini (NOAA) (Squirrel + PMF)
    • Steve Brown (Sonoma Tech) (PMF only)

Related Links

Email announcing the Clinic (sent 30-Jan-2009)

Dear all,

We would like to announce a "clinic" for the analysis of high-resolution ToF-AMS data (also including unit-resolution analyses). The idea is to provide a forum to discuss the analysis of high-resolution spectra (PIKA) in detail, and also including other topics such as PMF, APES (elemental analysis) and also unit-resolution analysis (Squirrel) in a format for which we don't have time in the regular Users Meetings.

Doug Worsnop, Donna Sueper, James Allan, Manjula Canagaratna, Qi Zhang, Jesse Kroll, Pete DeCarlo, and Jose Jimenez will be there. Ingrid Ulbrich will be there at least during the PMF discussions. The format will be something like:

  • first day with presentations of new features in PIKA and overview of analysis
  • Users bring up problems or special cases they have encountered in their work
  • Next several days: we intermittently break up and everyone works in their own data and we discuss interesting questions or problems as a group. We will also cover the other topics (PMF, APES, Squirrel) as appropriate, in a flexible way decided by consensus of the attendees.

The people who should attend are students, postdocs, and researchers who are currently actively analyzing high-resolution (and also unit-resolution) AMS data.

The location will be the University of Colorado-Boulder, and the dates will be Sat June 6 to Wed June 10, 2009. (Doug will only be present Sat to Mon, all the other organizers should be there the whole time). The dates are set because of Doug's other commitments.

Please let us know of your intention to attend, we are imagining about 25-30 people in attendance. Let us know if you have any suggestions about the content or the format.

Thanks,

-Jose et al.

Retrieved from "http://cires1.colorado.edu/jimenez-group/wiki/index.php?title=HR_Clinic&oldid=3722"