NeuroGAP seeks to build collaboration across Africa, particularly across early career researchers. Phenotyping working group has members across Kenya, Uganda, Ethiopia and South Africa, as well as from the Broad. Lots of clinicians, valuable for building phenotypes #WCPG2022
8 phenotyping manuscripts accepted, 8 more to come, describing the details of psychiatric phenotyping within and across different African countries #WCPG2022
See different loadings and importance of items in scales across different countries. Underlying cultural nuances in phenotyping that need to be captured in instruments #WCPG2022
[I'm not detailing the nuances of this work that well - papers will offer much better insight, e.g pubmed.ncbi.nlm.nih.gov/36028013/ ] #WCPG2022
Next up is Tim Bigdeli talking about multi-ancestry studies of schizophrenia and bipolar disorder in the Million Veteran Program #WCPG2022
Aim to ID 10k deeply characterised SCZ and BIP patients, with 4 hour sessions gathering information on these individuals #WCPG2022
After a brief hiatus due to phone issues, I'm back - @loldeloo is now speaking about the genetics of severe mental illness in South America, particularly Colombia #WCPG2022
The population history of the Americas has created a complex admixture. Here, focussing on admixed populations in Brazil and Colombia. #WCPG2022
Brazil has broad three way admixture, driven by Portuguese male and Amerindian female mating. Also large African population following enforced displacement (i.e. slave trade). Create one of the genetically heterogeneous countries in the world #WCPG2022
PUMAS Brazil aims to recruit 20k participants of Black/Pardos (mixed) ancestry. 10k cases with structured interview and 10K controls. #WCPG2022
PUMAS Colombia recruiting 1k patients from Cartagena, which also has three way admixture #WCPG2022
Paisa Colombia Biobank - Misión Origen. 9M individuals in mountain region of Colombia. Admixed populations from Spanish and Native American populations, multiple bottlenecks from mountain migrations. Results in extensive homozygosity and IBD segments (> Finnish) #WCPG2022
Biobank focusses on 50k participants with severe mental illness and 50k controls (from local healthcare provider). EHR phenotypes, available on 400k patients. First EHR linked biobank in South America #WCPG2022
Extracting data from EHR. Compared diagnoses to SCID interview - good concordance, kappa 0.7-0.9. A lot of non-concordance is BIP-MDD mismatches, many of which are intermediate (i.e. diagnostically challenging individuals, categorisation is difficult) #WCPG2022
Geospatial analysis shows distinct diagnosis specific hotspots. Also see disease trajectories: medrxiv.org/content/10.110… (more from first author later!) #WCPG2022
Recruitment across studies ongoing - lifting off after pandemic, hopeful for successful recruitment on time #WCPG2022
Phenotype harmonisation is a priority across the studies. Cross cultural training aiming for uniform administration of scales. Item-level harmonisation as well #WCPG2022
Now @danhowrigan will talk about Blended Genome-Exome. To capture genome diversity, could do deep WGS but this is expensive. Alternatively can do deep WES and GWAS array imputation. But most arrays and imputation are EUR biased. Logistical challenges as well #WCPG2022
Enter the Blended Genome Exome. Unbiased common variant capture. 1x WGS + WES. $150 per sample, less at scale. 60 samples per lane of sequencing. Lots of lessons learned from mass sequencing during COVID. #WCPG2022
Low pass imputation using GLIMPSE software. Low pass seq - 0-2 reads in a many places. Select haplotypes and impute likelihood through hidden Markov chain. #WCPG2022
How good is BGE? Compare to 30x as truth, and GSA chip imputation as competition. Needed to calibrate - maximum return for effort. #WCPG2022
BGE has more (11.7 Vs 6.9M) and more accurate SNP results for AFR than GSA. Need to do some mild QC (posterior > 0.8) to ensure high concordance with 30x #WCPG2022
1KG imputation better than HRC - 3M more SNPs. Capture best at high freq MAF > 5%, but still get >90% concordance even for singletons #WCPG2022
Imputing rare coding variants - capturing about 61% of coding singletons. Capture more from the deep WES. #WCPG2022
Imputing across ancestry subsets is generally high quality as well. Capturing a lot of SNPs. Runtime of imputation is robust to sample size - time per sample is stable as sample sizes increase #WCPG2022
• • •
Missing some Tweet in this thread? You can try to
force a refresh
