no-intercept produces h2=4.84... about ldsc HOT 8 CLOSED

Could you post the log file?

On Jun 23, 2015, at 6:05 PM, Roberto Toro [email protected] wrote:

Hello,
I'm analysing a quantitative trait (brain volume) in a population of unrelated subjects, controlled for population stratification, without genomic control. If I don't constraint the intercept, I obtain:
h2=0.1743 (0.0527)
but if I add the --no-intercept flag, then
h2=4.8404 (0.0757) !
what can be the reason?
thank you in advance!

—
Reply to this email directly or view it on GitHub.

from ldsc.

here it is:

• LD Score Regression (LDSC)
• Version 1.0.0
• (C) 2014-2015 Brendan Bulik-Sullivan and Hilary Finucane
• Broad Institute of MIT and Harvard / MIT Department of Mathematics
• GNU General Public License v3

  ---

Options:
--h2 ICV.sumstats.gz
--ref-ld-chr LDSCORE/baseline.
--out /Users/roberto/Desktop/ICV
--no-intercept
--w-ld-chr LDSCORE/weights.

Beginning analysis at Tue Jun 23 18:59:28 2015
Reading summary statistics from ICV.sumstats.gz ...
Read summary statistics for 1068341 SNPs.
Reading reference panel LD Score from LDSCORE/baseline.[1-22] ...
Read reference panel LD Scores for 1189907 SNPs.
Reading regression weight LD Score from LDSCORE/weights.[1-22] ...
Read regression weight LD Scores for 1242190 SNPs.
After merging with reference panel LD, 1066566 SNPs remain.
After merging with regression SNP LD, 1063925 SNPs remain.
Removed 0 SNPs with chi^2 > 80 (1063925 SNPs remain)
Total Observed scale h2: 4.8408 (0.0757)
Categories: base_0 Coding_UCSC_0 Coding_UCSC.extend.500_0 Conserved_LindbladToh_0 Conserved_LindbladToh.extend.500_0 CTCF_Hoffman_0 CTCF_Hoffman.extend.500_0 DGF_ENCODE_0 DGF_ENCODE.extend.500_0 DHS_peaks_Trynka_0 DHS_Trynka_0 DHS_Trynka.extend.500_0 Enhancer_Andersson_0 Enhancer_Andersson.extend.500_0 Enhancer_Hoffman_0 Enhancer_Hoffman.extend.500_0 FetalDHS_Trynka_0 FetalDHS_Trynka.extend.500_0 H3K27ac_Hnisz_0 H3K27ac_Hnisz.extend.500_0 H3K27ac_PGC2_0 H3K27ac_PGC2.extend.500_0 H3K4me1_peaks_Trynka_0 H3K4me1_Trynka_0 H3K4me1_Trynka.extend.500_0 H3K4me3_peaks_Trynka_0 H3K4me3_Trynka_0 H3K4me3_Trynka.extend.500_0 H3K9ac_peaks_Trynka_0 H3K9ac_Trynka_0 H3K9ac_Trynka.extend.500_0 Intron_UCSC_0 Intron_UCSC.extend.500_0 PromoterFlanking_Hoffman_0 PromoterFlanking_Hoffman.extend.500_0 Promoter_UCSC_0 Promoter_UCSC.extend.500_0 Repressed_Hoffman_0 Repressed_Hoffman.extend.500_0 SuperEnhancer_Hnisz_0 SuperEnhancer_Hnisz.extend.500_0 TFBS_ENCODE_0 TFBS_ENCODE.extend.500_0 Transcribed_Hoffman_0 Transcribed_Hoffman.extend.500_0 TSS_Hoffman_0 TSS_Hoffman.extend.500_0 UTR_3_UCSC_0 UTR_3_UCSC.extend.500_0 UTR_5_UCSC_0 UTR_5_UCSC.extend.500_0 WeakEnhancer_Hoffman_0 WeakEnhancer_Hoffman.extend.500_0
Observed scale h2: -1.3225 -0.0512 0.1766 0.3732 -0.0918 -0.2027 0.6231 -0.4405 0.9919 -0.1491 -0.1309 1.6781 -0.0649 0.1261 -0.0583 -0.4509 -0.0691 1.6653 0.8158 -0.9677 0.1593 -0.4617 0.3165 -1.0034 2.1976 0.0843 0.2893 -0.3945 0.1875 -0.8899 0.7366 -1.5568 1.4326 -0.1068 -0.0724 0.115 -0.0948 0.4398 0.1164 0.4274 -0.257 0.7446 -0.1407 -1.1031 0.8978 -0.1034 0.145 0.047 -0.0524 0.0414 -0.1579 0.1799 0.227
Observed scale h2 SE: 0.6792 0.0593 0.0981 0.0784 0.1786 0.0887 0.12 0.2334 0.3914 0.28 0.3406 0.4296 0.0384 0.056 0.1565 0.2525 0.2 0.2808 0.7109 0.7766 0.4399 0.4795 0.2398 0.5244 0.5317 0.1092 0.2069 0.2642 0.1088 0.2267 0.2697 1.8905 1.9415 0.0455 0.0684 0.1913 0.1988 0.4135 0.5323 1.0544 1.0665 0.2324 0.3375 0.274 0.5095 0.0842 0.1004 0.0431 0.0526 0.0253 0.0503 0.072 0.1564
Proportion of SNPs: 0.0886 0.0013 0.0057 0.0023 0.0295 0.0021 0.0063 0.0122 0.048 0.0099 0.0149 0.0442 0.0004 0.0017 0.0056 0.0136 0.0075 0.0252 0.0347 0.0374 0.0239 0.0298 0.0152 0.0378 0.054 0.0037 0.0118 0.0226 0.0034 0.0112 0.0204 0.0343 0.0352 0.0007 0.003 0.0028 0.0034 0.0409 0.0637 0.0149 0.0152 0.0117 0.0304 0.0306 0.0676 0.0016 0.0031 0.001 0.0024 0.0005 0.0025 0.0019 0.0079
Proportion of h2g: -0.2732 -0.0106 0.0365 0.0771 -0.019 -0.0419 0.1287 -0.091 0.2049 -0.0308 -0.027 0.3467 -0.0134 0.026 -0.012 -0.0932 -0.0143 0.344 0.1685 -0.1999 0.0329 -0.0954 0.0654 -0.2073 0.454 0.0174 0.0598 -0.0815 0.0387 -0.1838 0.1522 -0.3216 0.2959 -0.0221 -0.015 0.0238 -0.0196 0.0908 0.024 0.0883 -0.0531 0.1538 -0.0291 -0.2279 0.1855 -0.0214 0.0299 0.0097 -0.0108 0.0085 -0.0326 0.0372 0.0469
Enrichment: -3.0838 -8.1403 6.3791 33.3916 -0.6441 -19.8315 20.4461 -7.4657 4.2714 -3.1101 -1.8199 7.8455 -34.9244 15.4148 -2.1455 -6.8312 -1.902 13.625 4.8632 -5.3398 1.3781 -3.2036 4.3086 -5.4853 8.4123 4.7057 5.0607 -3.6004 11.2781 -16.4551 7.4493 -9.3693 8.4117 -29.5413 -5.0449 8.6061 -5.7226 2.2234 0.3775 5.9178 -3.492 13.1093 -0.9551 -7.4468 2.7437 -13.2383 9.7065 9.9235 -4.5408 17.7907 -13.2438 19.8895 5.9516
Coefficients: -2.2314e-07 -5.8902e-07 4.6158e-07 2.4162e-06 -4.6606e-08 -1.4350e-06 1.4795e-06 -5.4021e-07 3.0908e-07 -2.2504e-07 -1.3169e-07 5.6769e-07 -2.5271e-06 1.1154e-06 -1.5525e-07 -4.9430e-07 -1.3763e-07 9.8589e-07 3.5189e-07 -3.8638e-07 9.9721e-08 -2.3181e-07 3.1177e-07 -3.9691e-07 6.0871e-07 3.4050e-07 3.6619e-07 -2.6052e-07 8.1607e-07 -1.1907e-06 5.3902e-07 -6.7795e-07 6.0866e-07 -2.1376e-06 -3.6504e-07 6.2273e-07 -4.1408e-07 1.6088e-07 2.7316e-08 4.2821e-07 -2.5268e-07 9.4858e-07 -6.9110e-08 -5.3884e-07 1.9853e-07 -9.5790e-07 7.0235e-07 7.1805e-07 -3.2857e-07 1.2873e-06 -9.5831e-07 1.4392e-06 4.3065e-07
Coefficient SE: 1.1461e-07 6.8259e-07 2.5644e-07 5.0761e-07 9.0636e-08 6.2835e-07 2.8488e-07 2.8620e-07 1.2195e-07 4.2264e-07 3.4259e-07 1.4534e-07 1.4962e-06 4.9562e-07 4.1695e-07 2.7677e-07 3.9807e-07 1.6623e-07 3.0665e-07 3.1010e-07 2.7546e-07 2.4078e-07 2.3617e-07 2.0742e-07 1.4727e-07 4.4088e-07 2.6193e-07 1.7446e-07 4.7360e-07 3.0327e-07 1.9732e-07 8.2329e-07 8.2489e-07 9.1150e-07 3.4493e-07 1.0360e-06 8.6858e-07 1.5126e-07 1.2491e-07 1.0563e-06 1.0487e-06 2.9609e-07 1.6581e-07 1.3384e-07 1.1267e-07 7.8025e-07 4.8667e-07 6.5800e-07 3.2933e-07 7.8755e-07 3.0539e-07 5.7599e-07 2.9661e-07
Lambda GC: 1.0527
Mean Chi^2: 1.0576
Intercept: constrained to 1
Analysis finished at Tue Jun 23 19:00:29 2015
Total time elapsed: 1.0m:0.4s

from ldsc.

Could you also post the log file from the analysis where the heritability
estimate was smaller?

Also, what is your sample size?

Best,

Hilary

On Wed, Jun 24, 2015 at 11:50 AM, Roberto Toro [email protected]
wrote:

here it is:

• LD Score Regression (LDSC)
• Version 1.0.0
• (C) 2014-2015 Brendan Bulik-Sullivan and Hilary Finucane
• Broad Institute of MIT and Harvard / MIT Department of Mathematics
• GNU General Public License v3

  ---

Options:
--h2 ICV.sumstats.gz
--ref-ld-chr LDSCORE/baseline.
--out /Users/roberto/Desktop/ICV
--no-intercept
--w-ld-chr LDSCORE/weights.

Beginning analysis at Tue Jun 23 18:59:28 2015
Reading summary statistics from ICV.sumstats.gz ...
Read summary statistics for 1068341 SNPs.
Reading reference panel LD Score from LDSCORE/baseline.[1-22] ...
Read reference panel LD Scores for 1189907 SNPs.
Reading regression weight LD Score from LDSCORE/weights.[1-22] ...
Read regression weight LD Scores for 1242190 SNPs.
After merging with reference panel LD, 1066566 SNPs remain.
After merging with regression SNP LD, 1063925 SNPs remain.
Removed 0 SNPs with chi^2 > 80 (1063925 SNPs remain)
Total Observed scale h2: 4.8408 (0.0757)
Categories: base_0 Coding_UCSC_0 Coding_UCSC.extend.500_0
Conserved_LindbladToh_0 Conserved_LindbladToh.extend.500_0 CTCF_Hoffman_0
CTCF_Hoffman.extend.500_0 DGF_ENCODE_0 DGF_ENCODE.extend.500_0
DHS_peaks_Trynka_0 DHS_Trynka_0 DHS_Trynka.extend.500_0
Enhancer_Andersson_0 Enhancer_Andersson.extend.500_0 Enhancer_Hoffman_0
Enhancer_Hoffman.extend.500_0 FetalDHS_Trynka_0
FetalDHS_Trynka.extend.500_0 H3K27ac_Hnisz_0 H3K27ac_Hnisz.extend.500_0
H3K27ac_PGC2_0 H3K27ac_PGC2.extend.500_0 H3K4me1_peaks_Trynka_0
H3K4me1_Trynka_0 H3K4me1_Trynka.extend.500_0 H3K4me3_peaks_Trynka_0
H3K4me3_Trynka_0 H3K4me3_Trynka.extend.500_0 H3K9ac_peaks_Trynka_0
H3K9ac_Trynka_0 H3K9ac_Trynka.extend.500_0 Intron_UCSC_0
Intron_UCSC.extend.500_0 PromoterFlanking_Hoffman_0
PromoterFlanking_Hoffman.extend.500_0 Promoter_UCSC_0
Promoter_UCSC.extend.500_0 Repressed_Hoffman_0
Repressed_Hoffman.extend.500_0 SuperEnhancer_Hnisz_0
SuperEnhancer_Hnisz.extend.500_0 TFBS_ENCODE_0 TFBS_ENCODE.extend.500_0
Transcribed_Hoffman_0 Transcribed_Hoffman.extend.500_0 TSS_Hoffman_0
TSS_Hoffman.extend.500_0 UTR_3_UCSC_0 UTR_3_UCSC.extend.500_0 UTR_5_UCSC_0
UTR_5_UCSC.extend.500_0 WeakEnhancer_Hoffman_0
WeakEnhancer_Hoffman.extend.500_0
Observed scale h2: -1.3225 -0.0512 0.1766 0.3732 -0.0918 -0.2027 0.6231
-0.4405 0.9919 -0.1491 -0.1309 1.6781 -0.0649 0.1261 -0.0583 -0.4509
-0.0691 1.6653 0.8158 -0.9677 0.1593 -0.4617 0.3165 -1.0034 2.1976 0.0843
0.2893 -0.3945 0.1875 -0.8899 0.7366 -1.5568 1.4326 -0.1068 -0.0724 0.115
-0.0948 0.4398 0.1164 0.4274 -0.257 0.7446 -0.1407 -1.1031 0.8978 -0.1034
0.145 0.047 -0.0524 0.0414 -0.1579 0.1799 0.227
Observed scale h2 SE: 0.6792 0.0593 0.0981 0.0784 0.1786 0.0887 0.12
0.2334 0.3914 0.28 0.3406 0.4296 0.0384 0.056 0.1565 0.2525 0.2 0.2808
0.7109 0.7766 0.4399 0.4795 0.2398 0.5244 0.5317 0.1092 0.2069 0.2642
0.1088 0.2267 0.2697 1.8905 1.9415 0.0455 0.0684 0.1913 0.1988 0.4135
0.5323 1.0544 1.0665 0.2324 0.3375 0.274 0.5095 0.0842 0.1004 0.0431 0.0526
0.0253 0.0503 0.072 0.1564
Proportion of SNPs: 0.0886 0.0013 0.0057 0.0023 0.0295 0.0021 0.0063
0.0122 0.048 0.0099 0.0149 0.0442 0.0004 0.0017 0.0056 0.0136 0.0075 0.0252
0.0347 0.0374 0.0239 0.0298 0.0152 0.0378 0.054 0.0037 0.0118 0.0226 0.0034
0.0112 0.0204 0.0343 0.0352 0.0007 0.003 0.0028 0.0034 0.0409 0.0637 0.0149
0.0152 0.0117 0.0304 0.0306 0.0676 0.0016 0.0031 0.001 0.0024 0.0005 0.0025
0.0019 0.0079
Proportion of h2g: -0.2732 -0.0106 0.0365 0.0771 -0.019 -0.0419 0.1287
-0.091 0.2049 -0.0308 -0.027 0.3467 -0.0134 0.026 -0.012 -0.0932 -0.0143
0.344 0.1685 -0.1999 0.0329 -0.0954 0.0654 -0.2073 0.454 0.0174 0.0598
-0.0815 0.0387 -0.1838 0.1522 -0.3216 0.2959 -0.0221 -0.015 0.0238 -0.0196
0.0908 0.024 0.0883 -0.0531 0.1538 -0.0291 -0.2279 0.1855 -0.0214 0.0299
0.0097 -0.0108 0.0085 -0.0326 0.0372 0.0469
Enrichment: -3.0838 -8.1403 6.3791 33.3916 -0.6441 -19.8315 20.4461
-7.4657 4.2714 -3.1101 -1.8199 7.8455 -34.9244 15.4148 -2.1455 -6.8312
-1.902 13.625 4.8632 -5.3398 1.3781 -3.2036 4.3086 -5.4853 8.4123 4.7057
5.0607 -3.6004 11.2781 -16.4551 7.4493 -9.3693 8.4117 -29.5413 -5.0449
8.6061 -5.7226 2.2234 0.3775 5.9178 -3.492 13.1093 -0.9551 -7.4468 2.7437
-13.2383 9.7065 9.9235 -4.5408 17.7907 -13.2438 19.8895 5.9516
Coefficients: -2.2314e-07 -5.8902e-07 4.6158e-07 2.4162e-06 -4.6606e-08
-1.4350e-06 1.4795e-06 -5.4021e-07 3.0908e-07 -2.2504e-07 -1.3169e-07
5.6769e-07 -2.5271e-06 1.1154e-06 -1.5525e-07 -4.9430e-07 -1.3763e-07
9.8589e-07 3.5189e-07 -3.8638e-07 9.9721e-08 -2.3181e-07 3.1177e-07
-3.9691e-07 6.0871e-07 3.4050e-07 3.6619e-07 -2.6052e-07 8.1607e-07
-1.1907e-06 5.3902e-07 -6.7795e-07 6.0866e-07 -2.1376e-06 -3.6504e-07
6.2273e-07 -4.1408e-07 1.6088e-07 2.7316e-08 4.2821e-07 -2.5268e-07
9.4858e-07 -6.9110e-08 -5.3884e-07 1.9853e-07 -9.5790e-07 7.0235e-07
7.1805e-07 -3.2857e-07 1.2873e-06 -9.5831e-07 1.4392e-06 4.3065e-07
Coefficient SE: 1.1461e-07 6.8259e-07 2.5644e-07 5.0761e-07 9.0636e-08
6.2835e-07 2.8488e-07 2.8620e-07 1.2195e-07 4.2264e-07 3.4259e-07
1.4534e-07 1.4962e-06 4.9562e-07 4.1695e-07 2.7677e-07 3.9807e-07
1.6623e-07 3.0665e-07 3.1010e-07 2.7546e-07 2.4078e-07 2.3617e-07
2.0742e-07 1.4727e-07 4.4088e-07 2.6193e-07 1.7446e-07 4.7360e-07
3.0327e-07 1.9732e-07 8.2329e-07 8.2489e-07 9.1150e-07 3.4493e-07
1.0360e-06 8.6858e-07 1.5126e-07 1.2491e-07 1.0563e-06 1.0487e-06
2.9609e-07 1.6581e-07 1.3384e-07 1.1267e-07 7.8025e-07 4.8667e-07
6.5800e-07 3.2933e-07 7.8755e-07 3.0539e-07 5.7599e-07 2.9661e-07
Lambda GC: 1.0527
Mean Chi^2: 1.0576
Intercept: constrained to 1
Analysis finished at Tue Jun 23 19:00:29 2015
Total time elapsed: 1.0m:0.4s

—
Reply to this email directly or view it on GitHub
#35 (comment).

from ldsc.

Hi Hilary,
here's the log of the run without constrained intercept.
sample size: 11,373 subjects
thank you!
roberto
(also, 'ratio' is very high for ICV, 30%, however, I also obtain h24 for other phenotypes where ratio<20%)

• LD Score Regression (LDSC)
• Version 1.0.0
• (C) 2014-2015 Brendan Bulik-Sullivan and Hilary Finucane
• Broad Institute of MIT and Harvard / MIT Department of Mathematics
• GNU General Public License v3

  ---

Options:
--h2 ICV.sumstats.gz
--ref-ld-chr LDSCORE/baseline.
--out /Users/roberto/Desktop/ICV
--w-ld-chr LDSCORE/weights.

Beginning analysis at Tue Jun 23 18:56:16 2015
Reading summary statistics from ICV.sumstats.gz ...
Read summary statistics for 1068341 SNPs.
Reading reference panel LD Score from LDSCORE/baseline.[1-22] ...
Read reference panel LD Scores for 1189907 SNPs.
Reading regression weight LD Score from LDSCORE/weights.[1-22] ...
Read regression weight LD Scores for 1242190 SNPs.
After merging with reference panel LD, 1066566 SNPs remain.
After merging with regression SNP LD, 1063925 SNPs remain.
Removed 0 SNPs with chi^2 > 80 (1063925 SNPs remain)
Total Observed scale h2: 0.1743 (0.0527)
Categories: base_0 Coding_UCSC_0 Coding_UCSC.extend.500_0 Conserved_LindbladToh_0 Conserved_LindbladToh.extend.500_0 CTCF_Hoffman_0 CTCF_Hoffman.extend.500_0 DGF_ENCODE_0 DGF_ENCODE.extend.500_0 DHS_peaks_Trynka_0 DHS_Trynka_0 DHS_Trynka.extend.500_0 Enhancer_Andersson_0 Enhancer_Andersson.extend.500_0 Enhancer_Hoffman_0 Enhancer_Hoffman.extend.500_0 FetalDHS_Trynka_0 FetalDHS_Trynka.extend.500_0 H3K27ac_Hnisz_0 H3K27ac_Hnisz.extend.500_0 H3K27ac_PGC2_0 H3K27ac_PGC2.extend.500_0 H3K4me1_peaks_Trynka_0 H3K4me1_Trynka_0 H3K4me1_Trynka.extend.500_0 H3K4me3_peaks_Trynka_0 H3K4me3_Trynka_0 H3K4me3_Trynka.extend.500_0 H3K9ac_peaks_Trynka_0 H3K9ac_Trynka_0 H3K9ac_Trynka.extend.500_0 Intron_UCSC_0 Intron_UCSC.extend.500_0 PromoterFlanking_Hoffman_0 PromoterFlanking_Hoffman.extend.500_0 Promoter_UCSC_0 Promoter_UCSC.extend.500_0 Repressed_Hoffman_0 Repressed_Hoffman.extend.500_0 SuperEnhancer_Hnisz_0 SuperEnhancer_Hnisz.extend.500_0 TFBS_ENCODE_0 TFBS_ENCODE.extend.500_0 Transcribed_Hoffman_0 Transcribed_Hoffman.extend.500_0 TSS_Hoffman_0 TSS_Hoffman.extend.500_0 UTR_3_UCSC_0 UTR_3_UCSC.extend.500_0 UTR_5_UCSC_0 UTR_5_UCSC.extend.500_0 WeakEnhancer_Hoffman_0 WeakEnhancer_Hoffman.extend.500_0
Observed scale h2: -0.1019 -0.0477 0.1277 0.1067 -0.0816 -0.1363 0.2713 -0.0062 0.3536 0.0616 -0.1858 0.1957 -0.0266 -0.0148 -0.0211 -0.2937 -0.173 -0.1533 0.1698 -0.2379 0.2707 -0.3084 0.2828 0.0225 -0.08 -0.1075 0.2689 -0.2008 0.0978 -0.3781 0.4716 0.7258 -0.7615 -0.0388 0.0513 0.0727 -0.0733 0.0459 0.0719 0.386 -0.3124 0.5836 -0.5835 0.126 -0.2301 0.0385 -0.0016 0.0104 -0.0499 0.0109 -0.0767 -0.0318 0.0649
Observed scale h2 SE: 0.2861 0.043 0.0669 0.0492 0.1028 0.0708 0.1019 0.1824 0.2843 0.2147 0.2671 0.3462 0.0304 0.0424 0.124 0.1894 0.1599 0.2311 0.625 0.6795 0.3655 0.3928 0.1959 0.372 0.3667 0.0833 0.1554 0.1771 0.0838 0.174 0.1983 1.279 1.307 0.0358 0.055 0.1469 0.154 0.2198 0.3235 0.8178 0.8312 0.1742 0.2457 0.1571 0.2766 0.0658 0.0759 0.0403 0.0396 0.0216 0.0377 0.0559 0.0981
Proportion of SNPs: 0.0886 0.0013 0.0057 0.0023 0.0295 0.0021 0.0063 0.0122 0.048 0.0099 0.0149 0.0442 0.0004 0.0017 0.0056 0.0136 0.0075 0.0252 0.0347 0.0374 0.0239 0.0298 0.0152 0.0378 0.054 0.0037 0.0118 0.0226 0.0034 0.0112 0.0204 0.0343 0.0352 0.0007 0.003 0.0028 0.0034 0.0409 0.0637 0.0149 0.0152 0.0117 0.0304 0.0306 0.0676 0.0016 0.0031 0.001 0.0024 0.0005 0.0025 0.0019 0.0079
Proportion of h2g: -0.5843 -0.2737 0.7326 0.6122 -0.4683 -0.7816 1.5562 -0.0354 2.0285 0.3533 -1.0659 1.1224 -0.1528 -0.0849 -0.1212 -1.6849 -0.9925 -0.8794 0.9743 -1.3647 1.5526 -1.7689 1.6222 0.1292 -0.4591 -0.6166 1.5424 -1.1517 0.5608 -2.1687 2.7054 4.1634 -4.3681 -0.2223 0.2945 0.4171 -0.4207 0.2631 0.4123 2.2143 -1.792 3.3475 -3.3473 0.7229 -1.3199 0.2209 -0.009 0.0595 -0.2863 0.0625 -0.4398 -0.1824 0.3721
Enrichment: -6.5956 -210.7377 128.0999 265.1679 -15.8969 -370.2442 247.2068 -2.9016 42.285 35.6804 -71.7216 25.4009 -397.769 -50.28 -21.605 -123.557 -132.1794 -34.8281 28.115 -36.4527 65.0384 -59.4223 106.8863 3.4191 -8.5072 -166.5447 130.6022 -50.8849 163.2844 -194.1202 132.441 121.2958 -124.16 -297.8092 99.3275 151.0758 -122.9295 6.4398 6.472 148.4129 -117.8779 285.2906 -110.016 23.6224 -19.5257 136.8614 -2.9197 60.7634 -120.0043 130.0739 -178.5434 -97.6117 47.218
Coefficients: -1.7187e-08 -5.4916e-07 3.3381e-07 6.9100e-07 -4.1425e-08 -9.6481e-07 6.4419e-07 -7.5612e-09 1.1019e-07 9.2979e-08 -1.8690e-07 6.6192e-08 -1.0365e-06 -1.3102e-07 -5.6300e-08 -3.2197e-07 -3.4444e-07 -9.0758e-08 7.3264e-08 -9.4991e-08 1.6948e-07 -1.5485e-07 2.7853e-07 8.9097e-09 -2.2169e-08 -4.3400e-07 3.4033e-07 -1.3260e-07 4.2550e-07 -5.0585e-07 3.4513e-07 3.1608e-07 -3.2355e-07 -7.7605e-07 2.5884e-07 3.9369e-07 -3.2034e-07 1.6781e-08 1.6865e-08 3.8675e-07 -3.0718e-07 7.4343e-07 -2.8669e-07 6.1557e-08 -5.0882e-08 3.5664e-07 -7.6084e-09 1.5834e-07 -3.1272e-07 3.3896e-07 -4.6526e-07 -2.5436e-07 1.2304e-07
Coefficient SE: 4.8268e-08 4.9468e-07 1.7489e-07 3.1840e-07 5.2170e-08 5.0154e-07 2.4193e-07 2.2372e-07 8.8580e-08 3.2415e-07 2.6869e-07 1.1712e-07 1.1838e-06 3.7496e-07 3.3034e-07 2.0758e-07 3.1836e-07 1.3681e-07 2.6961e-07 2.7131e-07 2.2888e-07 1.9725e-07 1.9291e-07 1.4714e-07 1.0157e-07 3.3631e-07 1.9666e-07 1.1699e-07 3.6457e-07 2.3277e-07 1.4513e-07 5.5699e-07 5.5531e-07 7.1628e-07 2.7743e-07 7.9559e-07 6.7264e-07 8.0424e-08 7.5921e-08 8.1931e-07 8.1724e-07 2.2196e-07 1.2069e-07 7.6765e-08 6.1171e-08 6.0979e-07 3.6773e-07 6.1534e-07 2.4824e-07 6.7338e-07 2.2868e-07 4.4752e-07 1.8598e-07
Lambda GC: 1.0527
Mean Chi^2: 1.0576
Intercept: 1.0175 (0.008)
Ratio: 0.3036 (0.138)
Analysis finished at Tue Jun 23 18:57:17 2015
Total time elapsed: 1.0m:0.6s

from ldsc.

Okay, a few more questions:

(1) It looks like you might be using an older version. Could you try git
pull and check that that doesn't fix the problem?
(2) Was there any custom array data in the dataset?
(3) Could you run the analysis with --ref-ld-chr weights. instead of
--ref-ld-chr baseline. and send me the log files with and w/o intercept?

Thanks,

Hilary

On Wed, Jun 24, 2015 at 4:15 PM, Roberto Toro [email protected]
wrote:

Hi Hilary,
here's the log of the run without constrained intercept.
sample size: 11,373 subjects
thank you!
roberto
(also, 'ratio' is very high for ICV, 30%, however, I also obtain h24 for

other phenotypes where ratio<20%)

• LD Score Regression (LDSC)
• Version 1.0.0
• (C) 2014-2015 Brendan Bulik-Sullivan and Hilary Finucane
• Broad Institute of MIT and Harvard / MIT Department of Mathematics
• GNU General Public License v3

  ---

Options:
--h2 ICV.sumstats.gz
--ref-ld-chr LDSCORE/baseline.
--out /Users/roberto/Desktop/ICV
--w-ld-chr LDSCORE/weights.

Beginning analysis at Tue Jun 23 18:56:16 2015
Reading summary statistics from ICV.sumstats.gz ...
Read summary statistics for 1068341 SNPs.
Reading reference panel LD Score from LDSCORE/baseline.[1-22] ...
Read reference panel LD Scores for 1189907 SNPs.
Reading regression weight LD Score from LDSCORE/weights.[1-22] ...
Read regression weight LD Scores for 1242190 SNPs.
After merging with reference panel LD, 1066566 SNPs remain.
After merging with regression SNP LD, 1063925 SNPs remain.
Removed 0 SNPs with chi^2 > 80 (1063925 SNPs remain)
Total Observed scale h2: 0.1743 (0.0527)
Categories: base_0 Coding_UCSC_0 Coding_UCSC.extend.500_0
Conserved_LindbladToh_0 Conserved_LindbladToh.extend.500_0 CTCF_Hoffman_0
CTCF_Hoffman.extend.500_0 DGF_ENCODE_0 DGF_ENCODE.extend.500_0
DHS_peaks_Trynka_0 DHS_Trynka_0 DHS_Trynka.extend.500_0
Enhancer_Andersson_0 Enhancer_Andersson.extend.500_0 Enhancer_Hoffman_0
Enhancer_Hoffman.extend.500_0 FetalDHS_Trynka_0
FetalDHS_Trynka.extend.500_0 H3K27ac_Hnisz_0 H3K27ac_Hnisz.extend.500_0
H3K27ac_PGC2_0 H3K27ac_PGC2.extend.500_0 H3K4me1_peaks_Trynka_0
H3K4me1_Trynka_0 H3K4me1_Trynka.extend.500_0 H3K4me3_peaks_Trynka_0
H3K4me3_Trynka_0 H3K4me3_Trynka.extend.500_0 H3K9ac_peaks_Trynka_0
H3K9ac_Trynka_0 H3K9ac_Trynka.extend.500_0 Intron_UCSC_0
Intron_UCSC.extend.500_0 PromoterFlanking_Hoffman_0
PromoterFlanking_Hoffman.extend.500_0 Promoter_UCSC_0
Promoter_UCSC.extend.500_0 Repressed_Hoffman_0
Repressed_Hoffman.extend.500_0 SuperEnhancer_Hnisz_0
SuperEnhancer_Hnisz.extend.500_0 TFBS_ENCODE_0 TFBS_ENCODE.extend.500_0
Transcribed_Hoffman_0 Transcribed_Hoffman.extend.500_0 TSS_Hoffman_0
TSS_Hoffman.extend.500_0 UTR_3_UCSC_0 UTR_3_UCSC.extend.500_0 UTR_5_UCSC_0
UTR_5_UCSC.extend.500_0 WeakEnhancer_Hoffman_0
WeakEnhancer_Hoffman.extend.500_0
Observed scale h2: -0.1019 -0.0477 0.1277 0.1067 -0.0816 -0.1363 0.2713
-0.0062 0.3536 0.0616 -0.1858 0.1957 -0.0266 -0.0148 -0.0211 -0.2937 -0.173
-0.1533 0.1698 -0.2379 0.2707 -0.3084 0.2828 0.0225 -0.08 -0.1075 0.2689
-0.2008 0.0978 -0.3781 0.4716 0.7258 -0.7615 -0.0388 0.0513 0.0727 -0.0733
0.0459 0.0719 0.386 -0.3124 0.5836 -0.5835 0.126 -0.2301 0.0385 -0.0016
0.0104 -0.0499 0.0109 -0.0767 -0.0318 0.0649
Observed scale h2 SE: 0.2861 0.043 0.0669 0.0492 0.1028 0.0708 0.1019
0.1824 0.2843 0.2147 0.2671 0.3462 0.0304 0.0424 0.124 0.1894 0.1599 0.2311
0.625 0.6795 0.3655 0.3928 0.1959 0.372 0.3667 0.0833 0.1554 0.1771 0.0838
0.174 0.1983 1.279 1.307 0.0358 0.055 0.1469 0.154 0.2198 0.3235 0.8178
0.8312 0.1742 0.2457 0.1571 0.2766 0.0658 0.0759 0.0403 0.0396 0.0216
0.0377 0.0559 0.0981
Proportion of SNPs: 0.0886 0.0013 0.0057 0.0023 0.0295 0.0021 0.0063
0.0122 0.048 0.0099 0.0149 0.0442 0.0004 0.0017 0.0056 0.0136 0.0075 0.0252
0.0347 0.0374 0.0239 0.0298 0.0152 0.0378 0.054 0.0037 0.0118 0.0226 0.0034
0.0112 0.0204 0.0343 0.0352 0.0007 0.003 0.0028 0.0034 0.0409 0.0637 0.0149
0.0152 0.0117 0.0304 0.0306 0.0676 0.0016 0.0031 0.001 0.0024 0.0005 0.0025
0.0019 0.0079
Proportion of h2g: -0.5843 -0.2737 0.7326 0.6122 -0.4683 -0.7816 1.5562
-0.0354 2.0285 0.3533 -1.0659 1.1224 -0.1528 -0.0849 -0.1212 -1.6849
-0.9925 -0.8794 0.9743 -1.3647 1.5526 -1.7689 1.6222 0.1292 -0.4591 -0.6166
1.5424 -1.1517 0.5608 -2.1687 2.7054 4.1634 -4.3681 -0.2223 0.2945 0.4171
-0.4207 0.2631 0.4123 2.2143 -1.792 3.3475 -3.3473 0.7229 -1.3199 0.2209
-0.009 0.0595 -0.2863 0.0625 -0.4398 -0.1824 0.3721
Enrichment: -6.5956 -210.7377 128.0999 265.1679 -15.8969 -370.2442
247.2068 -2.9016 42.285 35.6804 -71.7216 25.4009 -397.769 -50.28 -21.605
-123.557 -132.1794 -34.8281 28.115 -36.4527 65.0384 -59.4223 106.8863
3.4191 -8.5072 -166.5447 130.6022 -50.8849 163.2844 -194.1202 132.441
121.2958 -124.16 -297.8092 99.3275 151.0758 -122.9295 6.4398 6.472 148.4129
-117.8779 285.2906 -110.016 23.6224 -19.5257 136.8614 -2.9197 60.7634
-120.0043 130.0739 -178.5434 -97.6117 47.218
Coefficients: -1.7187e-08 -5.4916e-07 3.3381e-07 6.9100e-07 -4.1425e-08
-9.6481e-07 6.4419e-07 -7.5612e-09 1.1019e-07 9.2979e-08 -1.8690e-07
6.6192e-08 -1.0365e-06 -1.3102e-07 -5.6300e-08 -3.2197e-07 -3.4444e-07
-9.0758e-08 7.3264e-08 -9.4991e-08 1.6948e-07 -1.5485e-07 2.7853e-07
8.9097e-09 -2.2169e-08 -4.3400e-07 3.4033e-07 -1.3260e-07 4.2550e-07
-5.0585e-07 3.4513e-07 3.1608e-07 -3.2355e-07 -7.7605e-07 2.5884e-07
3.9369e-07 -3.2034e-07 1.6781e-08 1.6865e-08 3.8675e-07 -3.0718e-07
7.4343e-07 -2.8669e-07 6.1557e-08 -5.0882e-08 3.5664e-07 -7.6084e-09
1.5834e-07 -3.1272e-07 3.3896e-07 -4.6526e-07 -2.5436e-07 1.2304e-07
Coefficient SE: 4.8268e-08 4.9468e-07 1.7489e-07 3.1840e-07 5.2170e-08
5.0154e-07 2.4193e-07 2.2372e-07 8.8580e-08 3.2415e-07 2.6869e-07
1.1712e-07 1.1838e-06 3.7496e-07 3.3034e-07 2.0758e-07 3.1836e-07
1.3681e-07 2.6961e-07 2.7131e-07 2.2888e-07 1.9725e-07 1.9291e-07
1.4714e-07 1.0157e-07 3.3631e-07 1.9666e-07 1.1699e-07 3.6457e-07
2.3277e-07 1.4513e-07 5.5699e-07 5.5531e-07 7.1628e-07 2.7743e-07
7.9559e-07 6.7264e-07 8.0424e-08 7.5921e-08 8.1931e-07 8.1724e-07
2.2196e-07 1.2069e-07 7.6765e-08 6.1171e-08 6.0979e-07 3.6773e-07
6.1534e-07 2.4824e-07 6.7338e-07 2.2868e-07 4.4752e-07 1.8598e-07
Lambda GC: 1.0527
Mean Chi^2: 1.0576
Intercept: 1.0175 (0.008)
Ratio: 0.3036 (0.138)
Analysis finished at Tue Jun 23 18:57:17 2015
Total time elapsed: 1.0m:0.6s

—
Reply to this email directly or view it on GitHub
#35 (comment).

from ldsc.

(1) updating to the most recent version seems to solve it :D Now I get
"Total Observed scale h2: 0.2545 (0.0326)" with intercept constrained, and
"h2: 0.1743 (0.0527)" without intercept constrained.
(2) no
(3) running the analysis with --ref-ld-chr weights. throws an error:

Beginning analysis at Thu Jun 25 21:10:41 2015
Reading summary statistics from ICV.sumstats.gz ...
Read summary statistics for 1068341 SNPs.
Reading reference panel LD Score from
/Applications/_Geno/LDSCORE/weights.[1-22] ...
Read reference panel LD Scores for 1242190 SNPs.
Traceback (most recent call last):
File "/Applications/_Geno/ldsc/ldsc.py", line 630, in
sumstats.estimate_h2(args, log)
File "/Applications/_Geno/ldsc/ldscore/sumstats.py", line 260, in
estimate_h2
args, log, args.h2)
File "/Applications/_Geno/ldsc/ldscore/sumstats.py", line 239, in
_read_ld_sumstats
M_annot = _read_M(args, log, n_annot)
File "/Applications/_Geno/ldsc/ldscore/sumstats.py", line 116, in _read_M
_splitp(args.ref_ld_chr), _N_CHR, common=(not args.not_M_5_50))
File "/Applications/_Geno/ldsc/ldscore/parse.py", line 167, in M_fromlist
return np.hstack([M(fh, num, N, common) for fh in flist])
File "/Applications/_Geno/ldsc/ldscore/parse.py", line 158, in M
x = np.sum([parsefunc(sub_chr(fh, i) + suffix) for i in xrange(1, num +
1)], axis=0)
File "/Applications/_Geno/ldsc/ldscore/parse.py", line 152, in
parsefunc = lambda y: [float(z) for z in open(y,
'r').readline().split()]
IOError: [Errno 2] No such file or directory:
'/Applications/_Geno/LDSCORE/weights.1.l2.M_5_50'

I don't have any file called "weights.1.l2.M_5_50" in my LDSCORE directory.
Should I get a newer version of the weights as well?

thank you for your help!!

best,
roberto

On Thu, Jun 25, 2015 at 7:23 PM, hilaryfinucane [email protected]
wrote:

Okay, a few more questions:

(1) It looks like you might be using an older version. Could you try git
pull and check that that doesn't fix the problem?
(2) Was there any custom array data in the dataset?
(3) Could you run the analysis with --ref-ld-chr weights. instead of
--ref-ld-chr baseline. and send me the log files with and w/o intercept?

Thanks,

Hilary

from ldsc.

Ah oops, I forgot I hadn't included those files. If things are working now,
I'd just skip that extra debug step.

Best,

Hilary

On Thu, Jun 25, 2015 at 3:13 PM, Roberto Toro [email protected]
wrote:

(1) updating to the most recent version seems to solve it :D Now I get
"Total Observed scale h2: 0.2545 (0.0326)" with intercept constrained, and
"h2: 0.1743 (0.0527)" without intercept constrained.
(2) no
(3) running the analysis with --ref-ld-chr weights. throws an error:

Beginning analysis at Thu Jun 25 21:10:41 2015
Reading summary statistics from ICV.sumstats.gz ...
Read summary statistics for 1068341 SNPs.
Reading reference panel LD Score from
/Applications/_Geno/LDSCORE/weights.[1-22] ...
Read reference panel LD Scores for 1242190 SNPs.
Traceback (most recent call last):
File "/Applications/_Geno/ldsc/ldsc.py", line 630, in
sumstats.estimate_h2(args, log)
File "/Applications/_Geno/ldsc/ldscore/sumstats.py", line 260, in
estimate_h2
args, log, args.h2)
File "/Applications/_Geno/ldsc/ldscore/sumstats.py", line 239, in
_read_ld_sumstats
M_annot = _read_M(args, log, n_annot)
File "/Applications/_Geno/ldsc/ldscore/sumstats.py", line 116, in _read_M
_splitp(args.ref_ld_chr), _N_CHR, common=(not args.not_M_5_50))
File "/Applications/_Geno/ldsc/ldscore/parse.py", line 167, in M_fromlist
return np.hstack([M(fh, num, N, common) for fh in flist])
File "/Applications/_Geno/ldsc/ldscore/parse.py", line 158, in M
x = np.sum([parsefunc(sub_chr(fh, i) + suffix) for i in xrange(1, num +
1)], axis=0)
File "/Applications/_Geno/ldsc/ldscore/parse.py", line 152, in
parsefunc = lambda y: [float(z) for z in open(y,
'r').readline().split()]
IOError: [Errno 2] No such file or directory:
'/Applications/_Geno/LDSCORE/weights.1.l2.M_5_50'

I don't have any file called "weights.1.l2.M_5_50" in my LDSCORE directory.
Should I get a newer version of the weights as well?

thank you for your help!!

best,
roberto

On Thu, Jun 25, 2015 at 7:23 PM, hilaryfinucane [email protected]
wrote:

Okay, a few more questions:

(1) It looks like you might be using an older version. Could you try git
pull and check that that doesn't fix the problem?
(2) Was there any custom array data in the dataset?
(3) Could you run the analysis with --ref-ld-chr weights. instead of
--ref-ld-chr baseline. and send me the log files with and w/o intercept?

Thanks,

Hilary

—
Reply to this email directly or view it on GitHub
#35 (comment).

from ldsc.

great! thank you

On Thu, Jun 25, 2015 at 9:14 PM, hilaryfinucane [email protected]
wrote:

Ah oops, I forgot I hadn't included those files. If things are working now,
I'd just skip that extra debug step.

Best,

Hilary

On Thu, Jun 25, 2015 at 3:13 PM, Roberto Toro [email protected]
wrote:

(1) updating to the most recent version seems to solve it :D Now I get
"Total Observed scale h2: 0.2545 (0.0326)" with intercept constrained,
and
"h2: 0.1743 (0.0527)" without intercept constrained.
(2) no
(3) running the analysis with --ref-ld-chr weights. throws an error:

Beginning analysis at Thu Jun 25 21:10:41 2015
Reading summary statistics from ICV.sumstats.gz ...
Read summary statistics for 1068341 SNPs.
Reading reference panel LD Score from
/Applications/_Geno/LDSCORE/weights.[1-22] ...
Read reference panel LD Scores for 1242190 SNPs.
Traceback (most recent call last):
File "/Applications/_Geno/ldsc/ldsc.py", line 630, in
sumstats.estimate_h2(args, log)
File "/Applications/_Geno/ldsc/ldscore/sumstats.py", line 260, in
estimate_h2
args, log, args.h2)
File "/Applications/_Geno/ldsc/ldscore/sumstats.py", line 239, in
_read_ld_sumstats
M_annot = _read_M(args, log, n_annot)
File "/Applications/_Geno/ldsc/ldscore/sumstats.py", line 116, in _read_M
_splitp(args.ref_ld_chr), _N_CHR, common=(not args.not_M_5_50))
File "/Applications/_Geno/ldsc/ldscore/parse.py", line 167, in M_fromlist
return np.hstack([M(fh, num, N, common) for fh in flist])
File "/Applications/_Geno/ldsc/ldscore/parse.py", line 158, in M
x = np.sum([parsefunc(sub_chr(fh, i) + suffix) for i in xrange(1, num +
1)], axis=0)
File "/Applications/_Geno/ldsc/ldscore/parse.py", line 152, in
parsefunc = lambda y: [float(z) for z in open(y,
'r').readline().split()]
IOError: [Errno 2] No such file or directory:
'/Applications/_Geno/LDSCORE/weights.1.l2.M_5_50'

I don't have any file called "weights.1.l2.M_5_50" in my LDSCORE
directory.
Should I get a newer version of the weights as well?

thank you for your help!!

best,
roberto

On Thu, Jun 25, 2015 at 7:23 PM, hilaryfinucane <
[email protected]>
wrote:

Okay, a few more questions:

(1) It looks like you might be using an older version. Could you try
git
pull and check that that doesn't fix the problem?
(2) Was there any custom array data in the dataset?
(3) Could you run the analysis with --ref-ld-chr weights. instead of
--ref-ld-chr baseline. and send me the log files with and w/o
intercept?

Thanks,

Hilary

—
Reply to this email directly or view it on GitHub
#35 (comment).

—
Reply to this email directly or view it on GitHub
#35 (comment).

from ldsc.