All_PsychiatricDisorders_MetaAnalysis.txt.gz is the results from the cross-trait meta analysis that was conducted and described in "Exploring the genetic relationship between twelve psychiatric disorders" by Romero et al. in Nature Genetics. 
Meta-analysis was conducted using mvGWAMA, correcting for sample overlap. 
Psychiatric disorders include: attention deficit hyperactivity disorder, alcohol use disorder, anorexia nervosa, anxiety disorder, autism spectrum disorder, bipolar disorder, depression, insomnia disorder, obsessive-compulsive disorder, post-traumatic disorder, schizophrenia, and Tourette's syndrome. Original studies conducted within European ancestry (with the exception of schizophrenia; see supplementary material note 1). 
See primary publications for genotyping technology used. 
Sample size of cases is 453,848 and 1,245,648 for controls (total sample size: 1,699,495; not accounting for sample overlap). Both total and effective sample size per variant is given in the summary statistics.
 
ExcSCZ_PsychiatricDisorders_MetaAnalysis.txt.gz is the exact same analysis file only with the exclusion of the trait schizophrenia. 
Sample size of cases 386,458 and 1,151,633 for controls (total sample size: 1,538,090; not accounting for sample overlap). Both total and effective sample size per variant is given in the summary statistics.

Human genome reference build is GRCh37. 
Both files contains 3,851,466 variants, sorted by genomic location, and are described by the following columns:
chromosome = chromosome number
base_pair_location = base pair position in chromosome
effect_allele = effect allele
other_allele = non-effect allele
beta = effect size of variant
standard_error = standard error of effect size
effect_allele_frequency = allele frequency of effect allele
p_value = p-value of effect estimate
snp = uniq ID per variant, format -> CHR:BP:A1_A2
rsid = rs ID number of the SNP
n_sum = sample size by summing n per disorder GWAS
n_eff = effective sample size estimated in mvGWAMA
direction = directions of effect per cohort (order is alphabetical)

Additional comments for specific columns:
Beta, standard error and effect allele frequency are not part of the mvGWAMA output. We therefore added these columns to the output. 
effect_allele_frequency: all allele frequencies were imputed from the UKBiobank reference panel. 
Z-score transformation to beta estimates: $Z/sqrt((2*$MAF*(1-$MAF))*($Neff+ ($Z)^2))
Transformation to standard error: 1/sqrt((2*$MAF*(1-$MAF))*($Neff+($Z)^2))