Difference between differentialy expressed genes and violin plot

I am currently looking at differentialy expressed genes accross cluster.
In parallel, I have also looked at the expression of some specific genes accross cluster.
Here for “Retnla”, the violin plot suggest that this gene is deferentially expressed.

But when looking at DEG with those lines :


sc.tl.rank_genes_groups(sco, layer='normalised_cpm', groupby='cluster_renamed', method='wilcoxon', corr_method = 'benjamini-hochberg', pts = True)

logfc_threshold = 1
pval_threshold = 0.05
pct_threshold = 0.01

sc.tl.filter_rank_genes_groups(sco, groupby='cluster_renamed', key_added = 'filter_genes_rename', min_in_group_fraction=pct_threshold)

for cluster in clusters:
     filtered_results = sc.get.rank_genes_groups_df(sco, group=cluster, key='filter_genes_rename')
     filtered_results = filtered_results[(filtered_results['pvals'] < pval_threshold) & (filtered_results['logfoldchanges'] > logfc_threshold)]
     filtered_results = filtered_results[(filtered_results['pct in group'] > pct_threshold)]

     sc_filt.append(filtered_results)

df = pd.concat(sc_filt, ignore_index=True)

Retnla does not appear anywhere. Does anyone have an idea why ?