21 March 2014


bimodal and reproducible research

Next post Previous post

ddRADseq Dynabeads

ApTranscriptome

Yesterday struggled with finding that some ‘Bimodal’ transcripts did not in fact have bimodal expression. Found that the reason for this was in this code

exp_type = if(coef(lmout)['val'] > 0 & coef(lmout)['I(val^2)'] > 0) "High" else {
    if(coef(lmout)['val'] < 0 & coef(lmout)['I(val^2)'] < 0) "Low" else {
        if(coef(lmout)['val'] > 0 & coef(lmout)['I(val^2)'] < 0) "Intermediate" else {
            "convex"} # close else
        } # close else
    } # close else

For transcripts with overall high levels of expression, max expression at both “Low” and “High” temperatures were greater than twice the standard deviation of expression

if(coef(lmout)['val'] < 0 & coef(lmout)['I(val^2)'] < 0) "Low"

Fixed this by requiring that max expression at both “Low” and “High” temperatures was greater than twice the standard deviation of expression at the median temp (19.25C)

Number of transcripts with maximum expression at high, low, intermediate or both high and low (bimodal) temperatures.
  Bimodal High Intermediate Low NotExp
A22 1996 865 516 1913 290
Ar 1390 757 1508 1406 519

Correct

Number of transcripts with maximum expression at high, low, intermediate or both high and low (bimodal) temperatures.
  Bimodal High Intermediate Low NotExp
A22 353 941 516 3480 290
Ar 227 869 1508 2457 519

Reading

Archiving Reproducible Research with R and Dataverse

  • R package for achiving data and reports on Dataverse
  • looks promising…but how does it compare to FigShare??

Creative Commons Licence
This work is licensed under a Creative Commons Attribution 4.0 International License.