stanstrup / predret Goto Github PK

column type:

• RP
• HILIC
• Ion-exchange liquid chromatography (IELC)
• ion-pairing liquid chromatography (IPLC)
  
  
  
  

Mobile phase pH:

• acidic
• alkaline
• neutral
  
  
  
  

Mobile phase additive:

• Trifluoroacetic acid
• Formic acid
• Acetic acid
• Propionic acid
• Citrate
• Ammonium formate
• Ammonium acetate
• Ammonium bicarbonate
• Ammonium hydroxide
• Triethylamine

https://github.com/sjewo/shiny-loadbalancer

Seems it is loading too much info at startup somewhere. Need to be fixed. Takes too long to start the first time.

For some reason parLapply throws an error now in boot2ci.

Error in serialize(data, node$con) : error writing to connection

For the moment I am using the non-parallel version.

Steffen Neumann: Compound / name
Steffen Neumann: and RT / recorded_RT

• RT range
• mz range
• exact mass

Make errors red and warnings green.

There is no need to use 1000 points for plotting. It is too slow with the highchart.

Don't use values with higher than ??? error compared to general fit for prediction.

Right now stereo-chemistry is disregarded and the mean of the rts of isomers is used in the modelling step.
Thus the different in rt between diastereomers is disregarded.

The apache proxying of the shiny port doesn't work anymore with the newest shiny.

right now the statistics are based on the points used for the regression. That means that areas with few points count as much/little as points in areas with many points.
This is unfair as the points in scarcely populated areas often have large CI widths and new predictions in a random place will be affected mroe by this one point than any one point in populated areas.

Possible solution1:
weight the CIs by the width between the two points in each side.

Possible solution 2:
Make predictions with the models using many mono-spaced points to get unbiased CIs.

And in "Add or modify system " you could allow "select existing system" from somebody else, so I would use that as a template

Caused by showOutput when renderChart2 is set to NULL. rCharts bug.