Is this for differential expression or something else? Also, does each condition have a sample from each batch?

I think you are mixing up two concepts here. Removing batch effects means removing unwanted technical variation between samples so you could compare them within a single statistical/differential framework. That might or might not be possible here, we would need a detailed info on what the samples are, what the experimental groups and batch groups are.

Meta-analysis is intended to find consistent trend across multiple differential analysis results exactly to NOT have the need to put all data into a single testing framework as batch effects etc. might massively skew these results.

Please add some details.

I have the expression data file (counts) that has been obtained from RNA-seq public data. The samples are kernels of a plant species inoculated with fungal pathogen, that have biological replicates. The experimental groups are 4 different studies on GEO. I have downloaded RNA-seq data, have done quality control and mapped on reference genome. Now I want to combine these 4 studies that I have count file for them. How I can remove technical variation among these 4 studies and what packages for meta-analysis?