The common approach is adding a very small number to all gene expressions. Let's say you add 1 or 0.001 (or anything similar) to all the numbers you have in the table. Then you do not have division by zero anymore and you can express FC. Be aware you create so called pseudo-counts since you don't work with the original data anymore. 

fold change in this case does not have sense.

just consider activation or suppression of gene expression.

Thanks Christina but we need to know fold change for constructing any fig. like-volcano plot for differential gene expression or network analysis etc.

I have whole rna seq data, some are like this, other are up and down so i c.ant keep two criteria.

i have to go with one criteria

The common approach is adding a very small number to all gene expressions. Let's say you add 1 or 0.001 (or anything similar) to all the numbers you have in the table. Then you do not have division by zero anymore and you can express FC. Be aware you create so called pseudo-counts since you don't work with the original data anymore.