Provide a list of TitanCall abbreviations in an easily-accessible form

[lbeltrame]

As the subject says, having to hunt a supplementary table in a manuscript to see what an abbreviation indicates is awkward (and, IIRC, there is no "supplementary table 2" in the TitanCNA paper).

Ideally this should be in the README.md or in the repo (or in the help files).

[gavinha]

Hi Luca,

Sorry for the inconvenience. I was an oversight on my part not to have put this directly into the R documentation.
I will add these to the documentation. Also, I plan to port over some of the original website material (plus updates) into the GitHub repo wiki. I will try to get to this soon.
For now, the definitions are actually found in Supplementary Table 14 (sorry for the further confusion with the incorrect table number).

homozygous deletion (HOMD),  
hemizygous deletion LOH (DLOH),  
copy neutral LOH (NLOH),  
diploid heterozygous (HET),  
amplified LOH (ALOH),  
gain/duplication of 1 allele (GAIN),  
allele-specific copy number amplification (ASCNA),  
balanced copy number amplification (BCNA),  
unbalanced copy number amplification (UBCNA)

I typically use the Copy_Number, MajorCN, MinorCN columns in the *segs.txt files.

Best,
Gavin

[fpbarthel]

What are all the possible values for Corrected_Call, I've noticed that in addition to the 9 above there is also the HLAMP class? (not mentioned in Supp Table 14 either)

Also I've observed states > 20. I figured perhaps the table continues in a linear fashion, since there are 2 rows for CN=1, 3 for CN=2, 4 for CN=3, 5 for CN=4, 6 for CN=5. However, I've observed state 21 and 23 for CN=8?

I've copied Supp Table 14 below:

titan_state	titan_genotype	titan_total_copy_number	titan_call	titan_call_description
-1	NULL	NULL	OUT	Outlier state
0	NULL	0	HOMD	Homozygous deletion
1	A	1	DLOH	Hemizygous deletion LOH
2	B	1	DLOH	Hemizygous deletion LOH
3	AA	2	NLOH	Copy neutral LOH
4	AB	2	HET	Diploid heterozygous
5	BB	2	NLOH	Copy neutral LOH
6	AAA	3	ALOH	Amplified LOH
7	AAB	3	GAIN	Gain/duplication of 1 allele
8	ABB	3	GAIN	Gain/duplication of 1 allele
9	BBB	3	ALOH	Amplified LOH
10	AAAA	4	ALOH	Amplified LOH
11	AAAB	4	ASCNA	Allele-specific copy number amplification
12	AABB	4	BCNA	Balanced copy number amplification
13	ABBB	4	ASCNA	Allele-specific copy number amplification
14	BBBB	4	ALOH	Amplified LOH
15	AAAAA	5	ALOH	Amplified LOH
16	AAAAB	5	ASCNA	Allele-specific copy number amplification
17	AAABB	5	UBCNA	Unbalanced copy number amplification
18	AABBB	5	UBCNA	Unbalanced copy number amplification
19	ABBBB	5	ASCNA	Allele-specific copy number amplification
20	BBBBB	5	ALOH	Amplified LOH

Here are the (distinct) titan states from two of my TITAN seg output files, which seem to match each other, but not Supplementary Table 14?

TITAN_state	TITAN_call	Copy_Number
1	DLOH	1
2	NLOH	2
3	HET	2
4	ALOH	3
5	GAIN	3
6	ALOH	4
7	ASCNA	4
8	BCNA	4
9	ALOH	5
12	ALOH	6
21	ASCNA	8
23	UBCNA	8

TITAN_state	TITAN_call	Copy_Number
0	HOMD	0
1	DLOH	1
2	NLOH	2
3	HET	2
4	ALOH	3
5	GAIN	3
6	ALOH	4
7	ASCNA	4
8	BCNA	4
9	ALOH	5
10	ASCNA	5
11	UBCNA	5
13	ASCNA	6
14	UBCNA	6
15	BCNA	6
17	ASCNA	7
18	UBCNA	7
19	UBCNA	7
20	ALOH	8
21	ASCNA	8
22	UBCNA	8
23	UBCNA	8
24	BCNA	8

EDIT: looking at the code here it looks like this has to do with whether the parameter symmetric is set to TRUE or FALSE. I suppose looking at the two examples I shared above the maximum value is 24, so it's probably set to TRUE. Is this a new feature that was added after the paper was released?

[fpbarthel]

Working back from the code I reconstructed the following version of supplementary table 14 for when symmetric=TRUE. I'm not exactly sure I got everything correct, @gavinha could you verify?

It may be useful to add this to the wiki somewhere, where it is easily visible.

titan_state	titan_genotype	titan_total_copy_number	titan_call	titan_corrected_call	titan_call_description	titan_corrected_call_description
-1	NULL	NULL	OUT	OUT	Outlier state	Outlier state
0	NULL	0	HOMD	HOMD	Homozygous deletion	Homozygous deletion
1	A	1	DLOH	DLOH	Hemizygous deletion LOH	Hemizygous deletion LOH
2	AA	2	NLOH	NLOH	Copy neutral LOH	Copy neutral LOH
3	AB	2	HET	HET	Diploid heterozygous	Diploid heterozygous
4	AAA	3	ALOH	ALOH	Amplified LOH	Amplified LOH
5	AAB	3	GAIN	GAIN	Gain/duplication of 1 allele	Gain/duplication of 1 allele
6	AAAA	4	ALOH	ALOH	Amplified LOH	Amplified LOH
7	AAAB	4	ASCNA	ASCNA	Allele-specific copy number amplification	Allele-specific copy number amplification
8	AABB	4	BCNA	BCNA	Balanced copy number amplification	Balanced copy number amplification
9	AAAAA	5	ALOH	ALOH	Amplified LOH	Amplified LOH
10	AAAAB	5	ASCNA	ASCNA	Allele-specific copy number amplification	Allele-specific copy number amplification
11	AAABB	5	UBCNA	UBCNA	Unbalanced copy number amplification	Unbalanced copy number amplification
12	AAAAAA	6	ALOH	ALOH	Amplified LOH	Amplified LOH
13	AAAAAB	6	ASCNA	ASCNA	Allele-specific copy number amplification	Allele-specific copy number amplification
14	AAAABB	6	UBCNA	UBCNA	Unbalanced copy number amplification	Unbalanced copy number amplification
15	AAABBB	6	BCNA	BCNA	Balanced copy number amplification	Balanced copy number amplification
16	AAAAAAA	7	ALOH	ALOH	Amplified LOH	Amplified LOH
17	AAAAAAB	7	ASCNA	ASCNA	Allele-specific copy number amplification	Allele-specific copy number amplification
18	AAAAABB	7	UBCNA	UBCNA	Unbalanced copy number amplification	Unbalanced copy number amplification
19	AAAABBB	7	UBCNA	UBCNA	Unbalanced copy number amplification	Unbalanced copy number amplification
20	AAAAAAAA	8	ALOH	HLAMP	Amplified LOH	High-level amplification
21	AAAAAAAB	8	ASCNA	HLAMP	Allele-specific copy number amplification	High-level amplification
22	AAAAAABB	8	UBCNA	HLAMP	Unbalanced copy number amplification	High-level amplification
23	AAAAABBB	8	UBCNA	HLAMP	Unbalanced copy number amplification	High-level amplification
24	AAAABBBB	8	BCNA	HLAMP	Balanced copy number amplification	High-level amplification

[fpbarthel]

This is also not documented anywhere I believe, but males seem to get the NEUT corrected copy number designation on X-chromosomal events.

[gavinha]

Thanks @fpbarthel

As I probably mentioned in this issue, earlier, the table of states went down with the original website.

TitanCNA is usually run with symmetric=TRUE because it simplifies the model and has better runtime.

The chromosome X correction is a new addition and like many other things, I haven't had time to provide documentation. The idea with using NEUT by default for chrX is that it is neutral relative to the matched normal. Yes, there is only 1 copy originally, and any copy number changes in the signal should be scaled by half compared to autosomes (which originated with 2 copies). It's a subtle detail but I had to deal with this when analyzing prostate cancer data, so I thought I'd include it into the TitanCNA post-processing.

I just hope that it makes sense. Again for the actual formulation where this idea of using a baseline of 1 copy for chrX in males will lead to a scaling of 1/2, see the writeup here:
https://www.cell.com/cell/fulltext/S0092-8674(18)30648-2#secsectitle0075
under section **Copy number analysis - 10X Genomics WGS data" after step 6 of TITAN.

[fpbarthel]

Thank you for the clarification!

[sbamin]

@gavinha,

In ploidy 3 or more cases, I am seeing GAIN or ASCNA calls (not exceeding corrected copy of 4) in regions where canonical tumor suppressor like CDKN2A, PTEN are located. While other cases have expected HOMD and DLOH calls for these genes, I believe there is a possibility for GAIN/ASCNA regions to have an actual state of a deletion of one or more allele following whole genome duplication event. Comments?

I did enrichment analysis to test whether CDKN2A GAIN calls are in ploidy2 vs ploidy3 cases vs CDKN2A HOMD/DLOH calls (equally present in ploidy2 vs 3 cases). While is 18/19 GAIN calls (CN < 4) are in ploidy3 cases, I think I am missing something here because the way TITANCNA selects optimal solution, if a case had majority of GAIN (2+1) calls, it should fall under ploidy3 solution, and perhaps I need a different way to test significance, if any.

[gavinha]

Hi @sbamin

If regions are first deleted and then genome doubled, then (hopefully) the prediction is copy neutral LOH (NLOH) or amplified LOH (ALOH).

The tumor ploidy parameter represents, in practice, the average tumor copy number across the genome. So your interpretation is correct.

I am less clear what your question is. Do you have examples of plots that can help to illustrate it?

Thanks,
Gavin

[sbamin]

Thanks Gavin,

Indeed and early CDKN2A loss have LOH or NLOH for most cases. So, TitanCNA is calling it correctly. For CDKN2A calls with GAIN/ASCNA state, I am labelling them as late events if I've reasonable confidence for WGD based on TitanCNA major:minor CN of >5:<1, ploidy of >4, and segmental logR copy number of > 2 on at least 50% of of at least half of canonical chromosomes (last one is based on https://www.biorxiv.org/content/10.1101/415133v2). Ad-hoc rules for now and open for comments.

Samir