FARMS: A New Algorithm for Variable Selection

Table 1

Data description. Summary of cohort characteristics either joined or split by country (Lima and Durban) showing viral load and its transformation log10 distribution. Below, number of HLA alleles and OLP present in the database and frequency of the top 5 parameters for each category.

	Viral load			Log10 (Viral load)
	Global	Lima	Durban	Global	Lima	Durban
Median	37800	37240	37900	4.577	4.571	4.579
IQR	(8715; 131500)	(13310; 109000)	(7075; 138500)	(3.94; 5.119)	(4.124; 5.038)	(3.85; 5.141)
	HLA			OLP
	Global	Lima	Durban	Global	Lima	Durban
Total number	73	62	66	406	391	371
1st	C07	A02	B15	76	76	78
	(; 6.31%)	(; 12.80%)	(; 6.48%)	(; 2.77%)	(; 2.09%)	(; 3.37%)
2nd	A02	B35	C07	78	84	76
	(; 5.99%)	(; 7.54%)	(; 6.17%)	(; 2.69%)	(; 1.74%)	(; 3.12%)
3rd	B15	C04	A30	84	81	84
	(; 5.66%)	(; 6.70%)	(; 5.70%)	(; 2.48%)	(; 1.68%)	(; 2.86%)
4th	A30	C07	C06	25	85	25
	(; 4.58%)	(; 6.70%)	(; 5.39%)	(; 1.82%)	(; 1.46%)	(; 2.56%)
5th	C04	B39	A68	41	78	41
	(; 4.42%)	(; 4.49%)	(; 4.50%)	(; 1.80%)	(; 1.38%)	(; 2.17%)