Technical Skills Exercise

Below is a real world example of a problem that regularly occurs in the SomaLogic data science workflow.

Background

Every 6 months, samples from two subjects are run through the assay, with each sample run three times each (a total n = 6 runs). Predictions for the SomaScan Tests are then made for each of the 6 runs and the average for each subject/individual is calculated. This process is part of maintaining our regulatory certification. The final deliverable looks like this:

#   SubjectId  mean_testA  mean_testB
# 1         1        0.69    62.26667
# 2         2        0.63    61.74000

The current method for providing this final table involves a lot of cutting and pasting, because the files are disjoint. We want to move as far away as possible from any cut-and-paste operations, as they are very error prone. The information we need to calculate the table comes from two files

1. A predictions file Predictions.csv
2. An ID file Barcode.csv

Predictions.csv file:

In Predictions.csv, we have predictions for testA (probability) and testB (linear response).

• The first column contains a barcode that is tied back to an individual
• The second and third columns contain the predictions

#   barcode testA testB
# 1      d2  0.71 62.37
# 2      a4  0.63 62.72
# 3      y3  0.67 62.74
# 4      g6  0.69 61.69
# 5      w5  0.67 60.19
# 6      b1  0.59 62.31

Barcode.csv file:

This file allows us to link the barcodes in the predictions file to specific subjects.

• The first column contains the SubjectId
• The remaining columns contain the barcodes for the three replicate runs

#   SubjectId barcode_1 barcode_2 barcode_3
# 1         1        y3        d2        g6
# 2         2        a4        b1        w5

Instructions

1. clone this repository
2. create a branch for yourself
3. rename the tech-exercise.Rmd file to tech-exercise-<name>.Rmd
4. open tech-exercise-<name>.Rmd and follow the directions inside