In quevedor2/aneuploidy_score: Calculate the Aneuploidy Score from Chromosome Arm SCNAs/Aneuploidies

Overview

The point of this package is to calculate the Aneuploidy Score for a chromosomal arm SCNA/aneuploidy (CAA) as detailed in the following two papers:

• Shukla, A., Nguyen, T. H., Moka, S. B., Ellis, J. J., Grady, J. P., Oey, H., ... & Duijf, P. H. (2020). Chromosome arm aneuploidies shape tumour evolution and drug response. Nature communications, 11(1), 1-14.

• Cohen-Sharir, Y., McFarland, J. M., Abdusamad, M., Marquis, C., Bernhard, S. V., Kazachkova, M., ... & Ben-David, U. (2021). Aneuploidy renders cancer cells vulnerable to mitotic checkpoint inhibition. Nature, 1-6.

Details of Aneuploidy Score/CAA

Defintions:

• Aneuploidy score (AS): as described by Cohen-Sharir et al., is simply the total number of arm-level gains and losses for a tumor, adjusted for ploidy.

• Chromosomal arm-level SCNAs/Aneuploidies (CAA): as described by Shukla et al., are arm-level aneuploidies that are pervasive in cancer and found ~30x higher than expected based on the inverse-length distribution of focal SCNAs.

AS and CAAs are calculated two different ways:

• Shukla method:

  • A Log 2 Ratio (L2R)-based method
  • Identify gains and losses using a 0.2 and -0.2 threshold respectively
  • If the CN segment intersects the centromere positions, discard it (Line 118-126 (commit: 25ea7bd): https://github.com/pascalduijf/CAAs_1/blob/master/scripts/CAAs_from_cell_lines.py)
  • Flag the arm as a CAA if the total length for the gain/loss segment in the chromosome arm exceeds 90% (0.90) of the total chromosome arm
  • [Optional]: whole-genome doubling is determined if half of the autosomal genome had two or more copies of the more frequent (maternal or paternal) allele.

• Cohen-Sharir method:

  • A Total Copy Number (TCN)-based method
  • If a CN-segment spans the centromere, split it and assign the segment to its respective arm.
  • Take the segment-size weighted median total copy number of each chromosome arm (arm_wMedian)
  • Assign the arm_wMedian a loss/neutral/gain (-1/0/1) status based on relation to the cell lines ploidy
  • Using a 0.5 threshold (from the default round() function), determine whether the arm_wMedian is >, =, or < than the ploidy
  • AS = Total number of arm gains/losses

Demo/Guide

To have some data to work with, the CCLE 639-V cell line seg file is included in the package, as well as cytobands downloaded from UCSC table browser for:

• ucsc.[GENOMEBUILD].cytoband where GENOMEBUILD=hg19, hg38, mm10, mm9, mm39

library(AneuploidyScore)
data("seg")
data("ucsc.hg19.cytoband")

Where the seg has both total copy number (TCN) and L2R (seg.mean):

head(seg,3)

Using the UCSC cytoband information, we first clump the cytobands into p-arm, q-arm and centromere segments and split by chromosome:

cytoarm <- cytobandToArm(ucsc.hg19.cytoband)
head(cytoarm[[1]],3)

To use the Cohen-Sharir method, we select for the Total copy number column (TCN) in the seg file and estimate the total ploidy using the standard weighted-mean approach:

tcn_col <- 'TCN'
wgd_ploidy <- checkIfWGD(seg, tcn_col = tcn_col, threshold = 0.5,
                         wgd_gf = 0.5, ploidy_method = 'wmean')

While we used weighted mean in this section, ploidy can be estimated using either mean, median, weighted-mean, weighted-median, or a base-2-ploidy approach (mean, median, wmean, wmedian, multi_base2). The multi_base2 looks for the nearest base2 ploidy to the weighted-mean ploidy. For example:

• ploidy(wmean)=2.9, ploidy(multi_base2) = 2

• ploidy(wmean)=3, ploidy(multi_base2) = 4

• ploidy(wmean)=4.5, ploidy(multi_base2) = 4

• ploidy(wmean)=5, ploidy(multi_base2) = 6

Using the estimated ploidy, and a threshold of 0.5 (equivalent to the round() function), we identify the classification of Loss/Neutral/Gain (-1/0/1) for each CN segment (segCNclass) or the Cohen-Sharir's weighted-median chromosome arm (armCNclass).

threshold <- 0.5
seg_caa <- getCAA(seg, cytoarm, tcn_col=tcn_col, classifyCN=TRUE,
                  ploidy=wgd_ploidy['ploidy'], threshold=threshold)
head(seg_caa[[1]],3)

Finally, we reduce the CN-segment specific GRangesList object down to a simplified chromosome-arm representation and calculate the aneuploidy score:

caa <- reduceArms(seg_caa, caa_method=c('arm', 'seg'),
                  arm_ids = c('p', 'q'))
head(caa, 5)

AS <- colSums(abs(caa), na.rm = TRUE)
AS

quevedor2/aneuploidy_score documentation built on Feb. 26, 2021, 12:13 p.m.