General Meeting Presentation 1-29-14

2015-01-13

General Meeting Presentation 1-29-14
AbStat Presentation

Location of presentation file (new to old)
"C:\Users\kurtw_000\Documents\kurt\storage\CIM Research Folder\kwhittem\Presentations\2014\General Meeting Presentation Kurt Whittemore 2 1-29-14.pptx"
"C:\Users\kurtw_000\Documents\kurt\storage\CIM Research Folder\kwhittem\Presentations\2014\General Meeting Presentation Kurt Whittemore 1-29-14.pptx"

A Previous presentation to refer to
"C:\Users\kurtw_000\Documents\kurt\storage\CIM Research Folder\DR\2013\4-3-13\General Meeting Presentation 4-3-13\General Meeting Presentation 4-3-13.pptx"

Audio recording of presentation



Outline of presentation 


(maybe I can basically have one slide per section)


-Measuring humoral immune responses


-Measures that compose the AbStat


-Nature of entropy measure


-Entropy with previous biological data


-AbStat Applications


-array platforms


-AbStat changes with artificial antibody experiments


--Monoclonal affinities and AbStat


--mixing two monoclonal antibodies


--spiking antibody into sera


-Mouse vaccines and infections


--246 day timecourse


--Multiple mouse immunizations


-human vaccines


-Reduction in antibody repertoire complexity with lymphoma


-mouse cancer progression


-Human disease


--HT330k first chip disease dataset


--HT330k wafer 46


--CIM10k


--Alzheimer's disease


-Changes with age


--changes with age in mice


--changes with age in humans


-Rank of measures


-range of entropy


-Changes in entropy measure with removal of peptides


-Limitations of AbStat


-Value of AbStat compared to other classification methods


-Use and possible future applications


-Conclusion



post presentation comments 


post presentation comments


-try throwing out low intensitiy peptides as Chris suggested


-maybe I should just not show the antibody mixing experiment. . there was a lot of confusion and uncertainty about this experiment


-Chris mentioned Human Protein Atlas which has a lot of antibody and kd info


-Josh mentioned using the "Maximum Likelihood" technique to find the best parameters which distinguish the samples


-maybe 


-Chris thinks 3 monoclonal antibodies are too few to make any conclusions


-Neal was interested in trying different bin sizes or varying bin sizes based on the current level of surrounding noise


-Bart says it's not comparable to compare the antibodies when they are against different epitopes


-Chris mentioned showing the entropy value as a measure relative to the value of the maximum entropy


-Stephen mentioned during my presentation that it would have been better for me to use eastern blot instead of western blot on the first slide


--after reading the eastern blot wikipedia article, I don't see how this can be the case.  Eastern blots seem to be for post-translational modifications and actually have many non-accepted definitions (04-02-2014d1622).



email to Stephen after presentation
https://mail.google.com/mail/u/0/?ui=2&shva=1#search/stephen/143e076b0b43a5bf 


Hi Dr. Johnston,





From what I could tell from the presentation, here are the issues I could correct and things I could look into:





-Spend way less time talking on the entropy slide or don't even show the slide. . entropy is just an equation to measure disorder and I used it.





-Use fewer slides.





-Maybe I should show a machine learning table or two.  After talking with people, I was under the impression that they felt there was little difference between groups.  The machine learning results show that there certainly is a difference between groups, but they didn't see this.





-I think I should not show the antibody mixing experiment since I am not completely sure if there was just two antibodies or more, even after questioning people extensively about the experiment and looking at notebooks.  I stated this uncertainty in my dissertation





-People think 3 antibodies for monoclonal affinity data is too few, and that the antibody should be against the same epitope.  We don't have more data, so maybe this work will just have to be a start. .





-I could try throwing out the low intensity peptides as Chris suggested.





-I could try variable binning as Neal suggested.  Josh had some ideas about this.





-I could look at more cancer mouse data with Tiger.  He just sent me more data.





-I see no end to this. . . haha





If you have other corrections or improvements let me know.





Best,


Kurt



Talk with Andrey 


-maybe I could have another graphic on the history of entropy slide (maybe a narrow and broad histogram or the aaa abcd example)


-maybe I could have some pictures on the slide where I list all of the measures





Some slides