Credible set missing and variant duplicates

[deniseomahony01]

Hi Omer, I have used PolyFun to compute prior causal probabilities for functionally-informed fine-mapping on summary statistics data using UK-Biobank LD matrices and custom functional annotations. After running the finemapper.py script using FINEMAP and SuSiE, credible sets were constructed.

In a couple of regions however, some of the credible sets are missing, e.g. credible set '2' is missing from the output below. Note that this only happens for FINEMAP. When I fine-mapped the same regions using SuSiE no such issue occurred.

SNP	CHR	BP	A1	A2	PIP	BETA_MEAN	BETA_SD	DISTANCE_FROM_CENTER	CREDIBLE_SET
rs2981578	10	1.23E+08	C	T	1.00E+00	2.02E-02	1.84E-02	118734	1
rs2936871	10	1.23E+08	A	T	1.00E+00	2.02E-02	1.84E-02	134598	3
rs45631563	10	1.23E+08	A	T	1.00E+00	2.02E-02	1.84E-02	127747	4
rs7919434	10	1.23E+08	G	A	5.42E-01	1.12E-02	1.71E-02	135361	5
rs7919566	10	1.23E+08	G	A	3.12E-01	6.18E-03	1.38E-02	135412	5
rs4647913	10	1.23E+08	C	G	1.45E-01	2.78E-03	9.66E-03	135837	5
rs4752569	10	1.23E+08	A	T	1.00E+00	2.02E-02	1.84E-02	110113	6
rs4751843	10	1.23E+08	T	A	1.00E+00	2.02E-02	1.84E-02	109891	7
rs1047111	10	1.23E+08	C	T	1.00E+00	2.02E-02	1.84E-02	135984	8
rs1047111	10	1.23E+08	C	G	1.00E+00	2.02E-02	1.84E-02	135984	8
rs4751842	10	1.23E+08	A	T	1.00E+00	2.02E-02	1.84E-02	109890	9
rs4752568	10	1.23E+08	C	T	2.66E-01	1.15E-02	1.93E-02	108964	10
rs11199913	10	1.23E+08	A	T	2.39E-01	4.34E-03	7.81E-03	129191	10
rs4752537	10	1.23E+08	A	G	1.69E-01	3.05E-03	6.82E-03	128656	10
rs9420318	10	1.23E+08	G	A	1.55E-01	2.79E-03	6.55E-03	128395	10
rs10788169	10	1.23E+08	C	G	1.27E-01	-2.29E-03	6.05E-03	129189	10

A second issue I have noticed is that in some regions, the credible sets created contained variant duplicates (see example below). For example in the output below the variant rs142013821 appears both in Credible Set 2 and Credible Set 3. Note that this was true for both FINEMAP and SuSiE.

CHR	SNP	BP	A1	A2	SNPVAR	MAF	Z	N	P	PIP	BETA_MEAN	BETA_SD	DISTANCE_FROM_CENTER	CREDIBLE_SET
2	rs6721996	2.18E+08	A	G	8.12E-09	5.00E-01	-2.19E+01	245619	1.03E-106	1.00E+00	5.33E-02	1.97E-03	192138	1
2	rs142013821	2.18E+08	C	CTTAT	2.10E-08	3.22E-02	-7.05E+00	245619	1.85E-12	1.00E+00	-7.77E-01	3.42E-03	190412	2
2	rs142013821	2.18E+08	CTTAT	C	1.32E-08	4.06E-01	-9.91E+00	245619	3.85E-23	1.00E+00	-3.10E-01	2.79E-03	190412	3
2	rs7600412	2.18E+08	C	T	1.01E-07	3.60E-01	-1.32E+01	245619	1.08E-39	1.00E+00	4.59E-01	7.11E-03	183030	4
2	rs2372945	2.18E+08	C	A	1.32E-08	3.59E-01	-1.28E+01	245619	8.83E-38	1.00E+00	2.70E-01	6.83E-03	187489	5
2	rs113510705	2.18E+08	C	T	4.06E-08	2.93E-02	-3.25E+00	245619	1.15E-03	1.00E+00	-5.96E-02	1.97E-03	295071	6

Could you please share some thoughts/explanations on the above issues? Thank you.

[omerwe]

Hi @deniseomahony01,

Re the first problem: This sounds like a FINEMAP issue, so I'm afraid I can't help with that. You may want to reach out to the author of FINEMAP...

Re the second problem: These are in fact not duplicate SNPs. These are two duplicate SNPs that (unfortunately) share the same rsid. The first SNP represents an insertion event, and the second represents a deletion event (this can be seen in the fields A1 and A2).

Sorry I can't be of more help, and please let me know if there's something else I can help with.

[deniseomahony01]

Thank you for your response. I shall contact the relevant person. As for the second issue, apologies I didn't realise this was an indel. Thank you again.